real european low voltage test network for smart grid
TRANSCRIPT
Real European Low Voltage Test Network for
Smart Grid Applicationsby
Arpan Koirala
Submitted to the Department of Electrical Engineering, Electronics,Computers and Systems
in partial fulfillment of the requirements for the degree ofErasmus Mundus Master Course in Sustainable Transportation and
Electrical Power Systemsat the
UNIVERSIDAD DE OVIEDOAugust 2018
c© Universidad de Oviedo 2018. All rights reserved.
Author . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Certified by. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Pablo Arboleya
Associate ProfessorThesis Supervisor
Certified by. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .Lucia Suarez Ramon
Head of Infrastructure of the Remote Management System, EDPThesis Supervisor
Real European Low Voltage Test Network for Smart Grid
Applications
by
Arpan Koirala
Submitted to the Department of Electrical Engineering, Electronics, Computers andSystems
on August 22, 2018, in partial fulfillment of therequirements for the degree of
Erasmus Mundus Master Course in Sustainable Transportation and ElectricalPower Systems
Abstract
With the rise in smart grid implementation in Europe, the requirement of a real testsystem where different smart grid application can be tested is inevitable. Most ofthe existing test feeders or networks are based on North American style distributionsystem. There are very few test systems representing the European style but they arenot complete and do not represent actual European style network and feeder. In thisthesis, changes are proposed in the existing model for IEEE European LV test feederto make it represent an actual European style feeder. The proposed modified IEEEEuropean LV test feeder will have four-wires model instead of three. The Europeandistribution networks have TT grounding neutral, which is isolated from the ground ofconsumers. So the return path of all neutral current is through the neutral conductor,which gives rise in neutral voltage. The concept of neutral voltage is neglected in theoriginal test bus system which we have included in modified test feeder.
In the second part, a methodology is discussed to convert GIS data obtained fromEDP Spain to OpenDSS model using the set of codes. Further, a comprehensiveEuropean test network with 30 LV substations, 10290 buses and 8087 load pointwith smart meter data of 20 days has been proposed to represent the European styleLV distribution system. This test system provides the possibility to reconfigure LVnetwork as well as provides an important base for economic and technical studies inlarge scale Network.
Thesis Supervisor: Pablo ArboleyaTitle: Associate Professor
Thesis Supervisor: Lucia Suarez RamonTitle: Head of Infrastructure of the Remote Management System, EDP
3
Acknowledgments
I would like to thank Thesis supervisors Prof. Pablo Arboleya and Lucia Suarez
Ramon for their constant guidance during the process. I would also like to thank
EDP, Spain for providing the data used without which the thesis would have been
impossible.
I would like to thank EACEA and GSEP for the grants making me able to do my
graduate studies in three countries without any financial burden.
I would like to thank beautiful city of Gijon and all the people I meet here for
providing a good company for one year.
5
Contents
1 Introduction 13
1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.2 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.3 Research Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
1.4 Organization of Thesis . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2 State of Art 17
2.1 Distribution Test Feeder Models . . . . . . . . . . . . . . . . . . . . . 17
2.1.1 IEEE feeders . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.1.2 Other North American Test Circuits . . . . . . . . . . . . . . 22
2.1.3 Non-American Test and Representative Feeders . . . . . . . . 24
2.2 Analysis of Existing Models . . . . . . . . . . . . . . . . . . . . . . . 26
3 Simulation Software 29
3.1 OpenDSS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3.1.1 Power Flow Algorithm OpenDSS . . . . . . . . . . . . . . . . 30
3.1.2 OpenDSS Architecture . . . . . . . . . . . . . . . . . . . . . . 32
3.1.3 OpenDSS using COM Interface . . . . . . . . . . . . . . . . . 35
3.2 MATLAB R© . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
4 IEEE European Low Voltage Test Feeder 37
4.1 Description of the test feeder . . . . . . . . . . . . . . . . . . . . . . . 37
4.2 Limitation of the test feeder . . . . . . . . . . . . . . . . . . . . . . . 40
4.3 Modified IEEE European LV Test Feeder . . . . . . . . . . . . . . . . 43
4.3.1 Inclusion of Neutral Conductor TT Grounding . . . . . . . . . 43
6
4.3.2 Per Second Time-series Load Data . . . . . . . . . . . . . . . 51
5 Integration of Raw data to OpenDSS Code 53
5.1 Description of Raw Data . . . . . . . . . . . . . . . . . . . . . . . . . 54
5.1.1 Network Data . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
5.1.2 Smart Meter Record File . . . . . . . . . . . . . . . . . . . . . 58
5.1.3 Load Curve File . . . . . . . . . . . . . . . . . . . . . . . . . . 59
5.1.4 Phase Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
5.2 Indexing of Circuit Element Data . . . . . . . . . . . . . . . . . . . . 60
5.3 Complexities in Data . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
5.4 Algorithm to re-annotate the Bus, Line, and Load . . . . . . . . . . . 66
5.5 Program Structure for Data Extraction . . . . . . . . . . . . . . . . . 71
5.5.1 Network Data . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
5.5.2 Load Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
6 European Standard Distribution Network 75
6.1 Introduction to Test Network . . . . . . . . . . . . . . . . . . . . . . 76
6.2 Network Description . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
6.2.1 Use of MATLAB for Simulation . . . . . . . . . . . . . . . . . 85
7 Conclusions and Future Work 87
7.1 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
7.2 Future Works . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
A Tables 91
A.1 Resistance and Impedance of Cross Section . . . . . . . . . . . . . . . 91
A.2 Line Types and Line Code . . . . . . . . . . . . . . . . . . . . . . . . 92
B Figures 93
B.1 Isolated Element Report From OpenDSS . . . . . . . . . . . . . . . . 93
B.2 Example of Output by plot curve() . . . . . . . . . . . . . . . . . . 94
7
List of Figures
3-1 Building of Yprim for simple two phase Coupled Impedance . . . . . . 31
3-2 OpenDSS Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3-3 OpenDSS Interface with the master file . . . . . . . . . . . . . . . . . 34
4-1 One-line Diagram of European Low Voltage test feeder obtained from
the OpenDSS test case . . . . . . . . . . . . . . . . . . . . . . . . . . 38
4-2 Impedance modeling in IEEE European LV test feeder . . . . . . . . 43
4-3 Modified Impedance modeling as per real European style TT grounding 46
4-4 Phase B Voltage(Phase-ground) in all nodes in Original and Modified
European LV test Feeder . . . . . . . . . . . . . . . . . . . . . . . . . 48
4-5 Comparison of Neutral Voltage in modified European LV Test System
with the Total Power in the Transformer . . . . . . . . . . . . . . . . 49
4-6 Comparison of Current in different node for Line 170 between Modified
and Original European LV Test feeder . . . . . . . . . . . . . . . . . 50
4-7 Neutral Voltage profile along the feeder . . . . . . . . . . . . . . . . 50
4-8 Phase Current and Neutral Voltage at Bus 558 after Load modification
to Per-second load . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
5-1 Description of Raw file used to Create OpenDSS model of large
Distribution Network . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
5-2 File from GIS representing Transformer in Network . . . . . . . . . . 55
5-3 File Representing the LV circuit Breaker at the origin of feeder . . . . 56
5-4 File Representing the Line segments of each feeder . . . . . . . . . . . 56
5-5 File Representing the Coordinates of each Line segments of each feeder 57
5-6 File Representing the Load connection points in the network . . . . . 58
8
5-7 File Representing the position and status of Circuit Breaker in network 58
5-8 File Representing the Smart meter in the network . . . . . . . . . . . 59
5-9 File Representing the Load Curve . . . . . . . . . . . . . . . . . . . . 59
5-10 File Representing the Phase in which Load is connected . . . . . . . . 60
5-11 Simple Line Topology for Line Indexing . . . . . . . . . . . . . . . . . 62
5-12 Simple Line Topology for Bus Indexing . . . . . . . . . . . . . . . . . 62
5-13 Simple Line model to Explain the Data Complexity . . . . . . . . . . 64
5-14 Flowchart to link the circuit breaker with first point of line segment . 67
5-15 Algorithm for arranging Bus 1 and Bus 2 of Line Segment correctly . 68
6-1 Single line Diagram of Test Network . . . . . . . . . . . . . . . . . . 77
6-2 Output Structure of the Power Flow Simulation of Test Network . . 86
B-1 Isolated Circuit Element Report . . . . . . . . . . . . . . . . . . . . . 93
B-2 The Output file of Power Flow simulator given by plot curve(bus,2113,1,3) 94
9
List of Tables
2.1 Description of MV feeder subsets for Western Australian Prototype
feeder . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.2 Description of LV feeder subsets for Western Australian Prototype feeder 25
4.1 Definition of Line and Line code in IEEE European LV test feeder . . 40
4.2 Definition of Load and Load-shape in IEEE European LV test feeder 40
5.1 Line Indexing Codes . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
5.2 Bus Indexing Codes . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
5.3 Data of Line segment in raw form . . . . . . . . . . . . . . . . . . . 65
5.4 Working of Algorithm to check the position of Bus . . . . . . . . . . 69
A.1 Resistance and Impedance . . . . . . . . . . . . . . . . . . . . . . . . 91
A.2 Line Code for Different types of Lines used along with Circuit Breaker 92
10
Glossary
Acronyms
ADR Automated Demand Response.
COM Component Objective Model.
CTF Comprehensive Test Feeder.
DERs Distribution Energy Resources.
DG Distributed Generation.
DSAS Distribution System Analysis Subcommittee.
EPRI Electric Power Research Institute.
LV Low Voltage.
LVNTS Low Voltage Network Test System.
MV Medium Voltage.
NEV Neutral-to-Earth voltage.
NILM Non-Intrusive Load Monitoring.
PLC Powerline communication.
PNNL Pacific Northwest National Laboratory.
PV Photo-Voltaic.
TFWG Test Feeder Working Group.
Matrices
11
Yprim Primitive Admittance Matrix.
Ysys System Admittance Matrix.
Z012 Sequence Impedence Matrix.
Zabc Reduced Impedence Matrix.
Zprim Primitive Impedence Matrix.
Variables
Ipc Compensation Current from Power Conversion(PC) Element.
V Voltage.
12
Chapter 1
Introduction
1.1 Background
Distribution network is going through major transformation in modern time. With
the introduction of distributed energy resources (DERs) modern distribution networks
are forced to be converted into smart from the conventional one. This change has
promoted the introduction of various software and tools intended to develop advanced
control and distribution automation. The rise of smart meter penetration will increase
the need of the big data analytics and other computing resources to utilize the data
from smart meter for the benefit of utilities and consumer. The data from smart
meter will be a valuable resource in the study of load pattern, non-intrusive load
monitoring (NILM) and automated demand response (ADR).
With the transformation in the paradigm of the distribution system where the
consumers are no longer a consumer but ’prosumers’1 utilities should be equipped
with tools to monitor their prosumers in a more interactive way. Utilities in Europe
are spending their resources to convert all metering systems into smart metering
infrastructure. Still more has to be done in the field of utilizing the data from smart
meters for greater use apart from simple billing purpose. The basic hurdle in this
process is the shortage of test systems that are realistic enough and freely available
1As distributed resources are increasing, many author prefer to say ’prosumers’ to the consumerswho produce energy at some time and consume some other time
13
for testing various algorithms [1]. In recent years many test feeders for US-based
system have been released but very few feeders are available that represent the true
European LV test feeder. This restricts utilities from testing new algorithms and
control strategies which can have positive role in reducing cost for consumer and
utilities, increasing reliability of the supply, reduce the conflict of interest between
network providers, system operators and market platform, improve the regulations
for utility, better utilization of electrical assets and better integration of DERs and
ESDs [2].
Distribution networks are vital infrastructure and detailed load and customer data
raise the issue of privacy and security. Therefore only very few actual networks are
available for the public use as a test network. This has compelled us to use only the
available test feeders that have been published for some other reasons and sometimes
the purposes are far beyond the original intention of the creation of test feeder [1].
The other challenge with feeder models is that there are regional differences between
structure and operations of the circuit across the world. The electrical network is a
modern infrastructure built in the twentieth century. Although they were developed
parallel, there are some fundamental difference in North American and European
distribution network. A typical residential consumer in North America receives
three-phase power at 120/240 V at 60 Hz whereas in Europe the typical supply
for a consumer is 240/420 V at 50 Hz. Similarly, in terms of terminal network, the
North American system has a secondary transformer with a rating of between 10 kVA
and 50 kVA serving 1 to 10 consumers while in Europe the secondary transformers
are between 100 kVA to 1,000 kVA serving between 50 to 250 consumers through
secondary LV Networks. Variety of differences can be obtained in distribution feeders
within same country or in the same utility [3].
1.2 Motivation
The first set of openly available distribution test feeders was released in 1991 by IEEE
Test Feeder Working Group(TFWG) [4]. It had set of four radial feeders to provide
14
researchers with models to work in unbalanced load with untransposed distribution
network. All these models were based on North American distribution network. In
2015, for the first time, European style test case was created by TFWG. The feeder
was a simple representation of LV feeder in the United Kingdom but does not represent
the exact model of European style feeders.
EDP HC Energia in Asturias, Spain is a distribution utility responsible for more
than 650,000 customers in Spain 2. The company has selected the city Pola de Siero
in Asturias as InoVCity to implement their smart grid project 3. One of the objectives
of the project is to equip the whole city with smart metering infrastructure for the
active study of the load and make it test LV grid for research and development of
new technical solutions. With the availability of smart meter data, a suitable test
feeder is required to study the various aspects of smart grid features. The existing LV
test systems are not suitable due to various limitations. The communication network
for the smart grid implementation in Spain is power line communication (PLC).
The advantage of PLC is that it is integrated with grid infrastructure, low cost of
investment, no intervention from the third party and good performance [5]. But with
the bandwidth limitation imposed by this technology, it is hard to get per minute
load data from the smart meter using the PLC network [6] unlike the IEEE LV test
feeder load profile. This test feeder considers multiple grounded neutral in LV network
but in practice, the LV feeder in Spain has isolated neutral with neutral grounding
only at transformer end. This assumption limits the observation of neutral to ground
voltage parameter which is one of the important parameters for distribution networks
in Europe. Also, the loads assumed in European LV test feeder are all single-phase
load. But in real case, the loads are the mixture of single phase and three phase.
All these factors motivates to work towards the creation of new LV test network for
Spanish system.
2http://www.edphcenergia.es/es/mercado-y-zonas-de-distribucion/3http://www.edphcenergia.es/es/inovgrid/inovcity-pola-de-siero/
15
1.3 Research Objectives
The principal objective of this thesis is to analyze the existing IEEE European LV
test system and to propose a new one for the Spanish utility for their smart grid
applications. The ultimate aim is to produce a test network for EDP Spain in which
they can test various innovative ideas with the smart meter data. The first stage of
the research will be to convert the existing European Test feeder to four wires model
to see the change in the different system parameters. The second part will be to
create an OpenDSS model of two small feeders representing the typical urban and
rural feeders. The final aim will be to create a tool to convert GIS and smart meter
raw data to OpenDSS format using MATLAB so that the utility will easily be able
to convert any of their distribution feeders and networks to OpenDSS format.
1.4 Organization of Thesis
Chapter 1 presents the motivation behind the study and its possible applications
in EDP Spain.
Chapter 2 is the literature review on different available test feeders and networks
and smart grid applications researches that have been done by using them.
Chapter 3 deals about the simulation software used for simulation and its specific
features.
Chapter 4 is the description of the modification proposed in IEEE European
LV Test feeder to study the 4 wires distribution systems with TT grounded neutral
model and modification of load data to per-second time series data.
Chapter 5 has the analysis of raw GIS data of LV network and strategies used
to convert it into into OpenDSS script.
Chapter 6 describes the proposed real European LV test network of a complete
city for the study of various smart grid applications.
Chapter 7 is the conclusion of the master thesis and the future work that can
be done.
16
Chapter 2
State of Art
This chapter contains the description of existing test feeders that have been developed
by TFWG with analysis of their purposes and limitation. There is also a review of
the existing power flow solutions and motivation towards the selection of right one.
Finally, it also has a discussion regarding different software for power flow analysis in
distribution systems.
2.1 Distribution Test Feeder Models
A distribution network consists of the infrastructure required for delivery of electricity
from transmission/sub-transmission level to the final consumer. Test Feeders in
Distribution Networks are simplified models designed for specific purpose. The
straightforward way of creating the test system is by taking data of an actual feeder
from a real distribution network and removing the private data. The other method is
clustering several actual networks to build synthetic representative test systems. The
third method is building the test system through manual design. The fourth method
of building a test system is by using planner tools based on economic and technical
criteria to create a realistic distribution network [1].
17
2.1.1 IEEE feeders
In 1991, the first set of four test feeders was published by IEEE TFWG [4]. In
2001, a new feeder was added in test feeder list [7]. These test feeders were meant
for providing a solution benchmark for distribution system analysis tool. With
time the distribution system analysis tools have acquired sophisticated capabilities,
road-map to different test cases were proposed by the Distribution System Analysis
Subcommittee(DSAS) in 2009[8]. The paper explained the requirement of new
test-feeders for different seven topics: neutral-to-earth voltage(NEV), short circuit
benchmark model, DG protection, large distribution system model, inverter based
DG models, comprehensive test feeder and asymmetrical contingencies test feeder.
In 2008 line models for NEV analysis was presented in [9] based on the 4-wires multi-
grounded neutral distribution system proposed by Kersting [10]. In 2010, TFWG
proposed IEEE 8500-node test feeder in [11] to help benchmark the ability of the
software under test to handle larger circuits similar to those commonly used in
distribution system analysis and a Comprehensive Distribution Test Feeder in [12]
to test all standard components of a distribution system. In 2014, the 342-node Low
Voltage Networked Test System in [13] was designed to test the ability to study heavily
meshed and networked systems in urban areas with numerous parallel transformers
and parallel LV cables. The original intention and usage of these feeders are discussed
in section following:
13 Node Test Feeder
This is small and highly loaded test feeder designed in 1991 which includes
common features of North American distribution system like voltage regulators,
shunt capacitor bank, three phase and single phase overhead and underground lines,
and unbalanced loads [4]. This feeder provides a starting point to test convergence
problem for power-flow in a highly unbalanced system. This system provides a simple
system to check different algorithm for the unbalanced system, few of the highly
cited works are in optimal capacitor placement using genetic algorithm [14], control
18
of photo-voltaic with storage in micro-grid [15] and islanding detection in micro-grid
[16] which shows that the usage of this test bus is diverse and not limited to the initial
purpose to check convergence of the power flow solution.
123 Node Test Feeder
This test feeder presented in [4] operates at a nominal voltage of 4.16 kV. But it is
a good test feeder to study the problem of voltage drop that must be solved with
the application of voltage regulators and shunt capacitor [4]. This feeder is well-
behaved and has no convergence problem. The presence of four voltage regulators
makes it a good test case to assure coordinated regulation of voltage. It also
includes numbers of switches to enable simple testing of intra-feeder reconfiguration
strategy[1]. The most cited usage of the 123 node test system is optimal power flow
in unbalanced situation[17], comprehensive operational planning framework for self-
healing action[18] and stochastic reactive power management in smart grid[19] which
shows the wide range of its usage apart from the purpose it was defined for.
34 Node Test Feeder
This feeder described in [4] is a part of actual feeder in Arizona. The feeder is very
long and lightly loaded. Unbalanced load with both spot and distributed loads with
Shunt capacitors, two in-line voltage regulators, and one in-line transformer makes it
a typical feeder. Because of the long length and unbalanced loading, it can present
a convergence problem [1]. The most cited paper using this test bus was on optimal
DG placement [20] and the optimal allocation of energy storage system [21] apart
from the original purpose of the test feeder.
37 Node Test Feeder
This feeder was obtained from the actual feeder in California with all underground
network [4]. This test feeder was meant for testing the capability of software to handle
the three-wire delta system which is a less common system in use [7]. Very few works
has been done using this feeder, some of the representatives are on optimal power flow
19
with DERs [22], DG as a provider of reactive power [23] and small signal analysis of
micro-grid system [24].
4 Node Test Feeder
This feeder was published in [7] after ten years of original IEEE feeders. The primary
purpose of this feeder was to provide a simple system for testing simulations of all
possible three-phase transformer connections including open delta for both balanced
and unbalanced load. There were total 6 types of transformer connections for step-
down and step-up operations. This test feeder is widely used in testing different
algorithms for state estimation [25] or for testing of step-voltage regulators [26] apart
from the original purposes.
The NEV Test Feeder
The NEV Test Feeder presented in [9] is a medium size radial feeder with a base
voltage of 12.47 kV. In all previous methods, the assumption was made that there is
no impedance between the neutral conductor and the earth ground. This assumption
is the base for Kron’s reduction which reduces the primitive 4x4 impedance matrix
to 3x3 phase impedance matrix by assuming neutral voltage at point of grounding is
0.0 V. [27] This assumption is valid in many cases except the cases where the neutral
voltage rises above 0.0 V. Study of neutral voltage is an important issue when there
are problems with broken ground connection and harmonic studies. This test case is
widely used in load modeling [28] and harmonic analysis [29].
8500 Node test Feeder
This test feeder presented in [11] was designed to check the working of distribution
analysis tools for the larger network. It has incorporated the various features of
North American distribution system like multiple feeder regulators, center-tapped
transformers, per-phase capacitor bank and LV secondary distribution. The test case
has loads for both balanced and unbalanced condition. It is a 12.47 kV large radial
feeder with 170 km of overhead lines and underground cable. This test bus has been
20
used for different purposes like load modeling with time series load [30] and smart
grid and DERs integration studies [31].
Comprehensive Distribution Test Feeder(CTF)
This feeder presented in [12] was developed to test the models of all distribution
components and to test the convergence qualities of a verity of switching schemes.
To make it a complex test feeder, it includes three, two and single phase overhead
and underground lines with and without neutral, three phases, single phase and
center tapped transformers, step voltage regulators, induction machines, spot loads,
switched capacitor banks and single-phase 120 and 240 volt loads on center tapped
transformers. It also has modeling of the parallel underground and overhead lines.
Despite being complete with all type of available distribution elements, this test
feeder might give unrealistic result for studies of distribution system except the main
purpose of testing the distribution network simulation software [1]. This test feeder
has been used for some DG applications [32] apart from its original intention to test
software.
342 node Test system(LVNTS)
This test system defined in [13] was designed to present distribution system analysis
software with a heavily meshed and networked systems, systems with numerous
parallel transformers and modeling of low voltage parallel cables. This is called a
test system instead of test feeder as it has multiple 13.2 kV feeders supplying 120 V
highly meshed network and 480 V spot networks [3]. The system is in use in many
parts of US where numerous parallel lines are serving the consumer and fault in one
line isolates that line without affecting the consumers. This system has been used to
study optimal dispatch with DERs [33] and communication applications [34].
European(UK) Low Voltage Test Feeder
The European Low-Voltage Test Feeder is a 400 V radial system representing a type
of feeder in United Kingdom [35]. The voltage at the transformer end is kept at 416
21
V to comply with the typical operational values in the United Kingdom. It consists
of a single transformer and a single LV feeder with 55 single phase loads connected
in different phases. The load has time series data with 1440 values for each minute
of a day. This is the first IEEE test feeder operating at 50 Hz. The purpose of
this feeder was to accommodate the capability of software to operate in different
distribution systems around the globe [3]. This model is widely used in DERs and
smart grid applications [36, 37] and topology mapping [38] due to its time series data
and coordinates. The model still uses Carson’s equation with Kron’s reduction [27]
for converting the line into 3 wires model. Most of the European feeder is TT earthed,
i.e., the neutral is earthed at transformer end and the neutral provided to consumer
provides the return path to all neutral current. The application of Kron’s reduction
which assumes multiple neutral grounding performs well in the shorter LV network
but gives different result in a longer network with unbalanced load [39, 40]. This
master thesis proposes a modification in the IEEE LV test feeder to accommodate
the calculation of actual neutral voltage and is further discussed in Chapter 4.
Apart from these radial test feeders IEEE TFWG is now working on to create new
test cases on quasi-static time series, Chinese circuit, large-scale urban-core network
and microgrid[35]. It shows that TFWG is putting effort to prepare test feeder of
global systems but still it has to do a lot in the design of test feeders which are not
North American Type.
2.1.2 Other North American Test Circuits
There are different open source feeder models designed as a part of EPRI’s Green
Circuit project database and PNNL taxonomy feeder study. Unlike IEEE test feeders,
these feeders are not designed to check the power flow solving algorithm, but as a
representative feeder which can be used by researchers to use in case studies. EPRI
feeder set has six large feeders obtained from the real networks. All of them have data
of bus coordinate and time series data of load to provide a more realistic case. The
feeder J1, K1, and M1 are deeply focused on accessing the impact of distributed PV
and other DERs penetration. Remaining three feeders are for testing power flows and
22
other applications in smart grid environment. In 2009, PNNL also published a set
of 24 taxonomy radial distribution test feeders designed to represent US continental
area.
Feeder J1
This is a real feeder located in the North-eastern US available online at [41]. The
feeder has 1.7 MW of consumer-owned PV, 1300 residential, commercial and light
industrial customers and 58 miles of line. The complaint of the consumer regarding
high voltage was one of the reasons to choose it as a case feeder. The target of the
feeder was to verify if the PV and slow load tap changer is the cause of the over-
voltage. This feeder is used for voltage regulator applications in feeders with PV
[42].
Feeder K1
This is a real feeder located in the south eastern US [43]. The feeder has 1 MW
customer-owned PV along with 500 commercial and residential loads and 28 miles of
primary line. It has a substation load tap changer(LTC) and a single capacitor bank
at mid-feeder. The objective of this feeder is to study the power flow solution in high
PV penetration without voltage regulators.
Feeder M1
This is relatively shorter and compact circuit with 1500 consumers and 13 miles of
underground feeder [44]. This feeder has a unique radio-controlled capacitor bank
of rating 1200 kvar 3 phase. This test feeder aims at studying the various capacitor
bank management strategies.
Feeder Ckt5, Ckt7, and Ckt24
These are large-scale test networks presented in [45] available for testing power flow
in smart grids environment. Ckt5 has 1379 customers, Ckt7 has 5694 customers and
23
Ckt24 has 3885 customers. Ckt5 and Ckt7 are primarily residential with longer feeder
length while Ckt7 is shorter, compact and heavily loaded with commercial loads.
PNNL Taxonomy Feeders
The 24 radial distribution feeder models presented in [46] are the representative
synthetic feeders constructed based on the statistical analysis of distribution feeder
models supplied by 17 utilities of USA. The data of 575 distribution feeders from
151 separate substations were collected and a clustering algorithm was used to
generate synthetic test feeders. The original IEEE test feeders were meant for
checking the ability of software while they provide insufficient data relevant to
the actual distribution system. These feeders were designed to provide researchers
with the variety of feeder models in which smart grid ideas can be tested and
implemented. These test models were constructed assuming that distribution analysis
solution methods are valid. The US was divided into 5 climatic regions to carry out
the classification and 23 prototype feeders were generated based on the clustering
algorithm, to represent the respective subgroups. The 24th feeder was included to
represent a general feeder type that was found in all five of the climatic zone. These
feeders are used to study the various DERs and smart grid applications [47, 48].
2.1.3 Non-American Test and Representative Feeders
Several works have been done to create representative feeders in Europe, Australia,
and England where the feeders are different from the North American one.
Representative Western Australia Feeders
The Australian Representative feeder consists of nine prototypical MV feeders and
eight LV feeders as presented in [49]. The model uses the clustering algorithm as in
PNNL feeder[46] but with fewer numbers of variables. The prototype feeders were
selected from the analysis of 204 members of MV feeder database and 8858 members
of LV feeder database.
24
Table 2.1: Description of MV feeder subsets for Western Australian Prototype feeder[49]
Cluster number Description of feeder
1 light urban residential 12 heavy urban residential3 heavy mixed metropolitan4 moderate suburban residential5 light urban residential 26 commercial/industrial7 light suburban residential8 rural short9 rural long
Table 2.2: Description of LV feeder subsets for Western Australian Prototype feeder[49]
Cluster number Description of feeder
1 small commercial/residential with low transformer utilization and short lines2 residential mainly using overhead lines3 residential with high transformer utilization, mainly using underground lines4 residential with low transformer utilization, only using underground lines5 median mixed commercial/residential with mainly using overhead lines6 medium commercial/industrial with short LV lines7 small residential with high transformer utilization, using short total LV lines8 residential with high transformer utilization mainly using overhead lines
Table 2.1 and table 2.2 show the description of the type of the MV and LV feeders
selected as prototype feeders. Out of 9 MV feeders, 5 are residential feeders,2 are
rural feeders, 1 is commercial and 1 is a mixed metropolitan feeder. In LV feeders,
out of 8 feeders, 5 are residential and 3 are mixed feeders. These feeders are used
in mainly clustering applications as it provides easier way to cluster the distribution
feeders.
Representative Residential LV Feeder for North England
The representative feeder for England was presented in [50] based on the clustering
algorithm. The feeders were characterized using 19 features which included time-series
data, presence of DERs, network parameters and consumer classification. A set of 383
feeders were taken initially which was reduced to 232 after cleansing and validation.
25
In the paper, 11 representative feeders were chosen to represent the Distribution
system of North England, of which 8 were without PV and 3 had PV. These feeders
are basically used to study the impact of PV penetration [50, 51].
European Representative Electricity Distribution Network
This is the first complete European style reference network model that can automate
the layout of large-scale distribution network based on the street map and consumer
information[52]. Three large-scale networks and six feeder type networks were built
based on the data from 79 large European System Operators(DSOs). The indicators
as submitted by DSOs in an online survey was used to create the proposed network.
The constructed representative networks are categorized into two groups: Large-
scale Networks and Feeder type Networks. Large scale networks are the networks
including high voltage(HV), medium voltage(MV) and LV including both LV and MV
consumers, MV-LV feeders and MV/LV substations. Feeder type networks are divided
into MV feeder networks and LV feeder networks. Even though they represent the
large scale networks, all networks are considered as three phase and balanced which is
not the real scenario. The Reference Network Model(RNM) used by the authors is a
very good model for Distribution planning and network growth strategies as discussed
in [2].
2.2 Analysis of Existing Models
This thesis proposes the creation of the first large scale real European test network
covering LV network of a whole city considering the three-phase unbalanced system.
On reviewing the previous works no such system is previously available for research
and study. In IEEE Distribution test feeders, there is a large scale network (8500
node system) [11], but it represents the North American type distribution system.
The IEEE European Low Voltage Test Feeder presented in [35] is first European type
feeder, but it lacks proper representation as it has only one LV feeder originating from
transformer and 55 connected loads which are not the usual case. Further, it neglects
26
the TT type of earthing in which the neutral is grounded at the transformer only and
consumer earth and neutral are isolated. The time series load is a good feature of this
test feeder but in practice getting 1-minute load pattern is very hard if PLC is used
for communication. Also, the LV networks in Europe has multiple LV feeders used
for back-feeding the consumers during transformer outages which can be represented
through large scale network only. The synthetic network presented in [52] represents
the large scale European feeder but it fails to address the issue of unbalanced three-
phase network. Further, the network model created is a synthetic one and is based
on the street map. The development of the actual distribution network is a long-term
process and has many redundancies in the network path due to various restrictions.
The actual network is never optimal as the network changes with load-growth and
the feeder does not follow the shortest path always. These representative feeders are
good tools to study the distribution resources optimization and distribution planning
in a new city but not a complete one to interpret the actual European LV network.
This thesis thus aims to start the creation of a complete real European test network
based on real network data of a whole city. It is done in different steps. At first, the
IEEE European LV Test feeder is modified to represent the TT neutral ground system
to compare the differences. Second, small feeders are simulated using the 4 wires LV
system. The third will be the development of the script to create an OpenDSS file
directly from the GIS raw file and fourth will be the creation of complete LV network
model using the script.
27
Chapter 3
Simulation Software
This chapter discusses the basics of OpenDSS software and MATLAB R©tools used for
simulation of the distribution system. OpenDSS is the software used for the power
flow simulation while MATLAB is used for pre- and post-processing the data.
3.1 OpenDSS
OpenDSS is an open source software developed by EPRI for simulating utility power
distribution systems1. The OpenDSS tool has been used since 1997 and supports all
frequency domain analysis commonly performed in the distribution systems. It also
supports the other new type of analysis designed to meet the future needs related
to smart grid, grid modernization, storage system and renewable energy integration
[53].
One important feature of OpenDSS is that it uses a text interface unlike its
counterpart software which uses graphical interfaces like CYME2 and Synergi3. Each
piece of equipment should be represented on an individual line in the text editor. It
might be a huge task to model a large distribution system but advantageous in the
case where the real data is obtained from GIS where we can automate the process to
1http://smartgrid.epri.com/SimulationTool.aspx2http://www.cyme.com/software/#cap3https://www.dnvgl.com/services/power-distribution-system-and-electrical-simulation-software-
synergi-electric-5005
29
create the OpenDSS model from GIS raw data which is one of the scopes of this thesis
and will be discussed in Chapter 5. This chapter discusses the different components
used in distribution system simulation using OpenDSS.
3.1.1 Power Flow Algorithm OpenDSS
The program has different built-in power flow solution mode of which we are interested
in following power flow mode:
• Snapshot mode
• Daily Power Flow mode
• Yearly Power Flow mode
Snapshot mode solves the power flow for a single instant while daily and yearly
power flow mode can be used for the time sequence data where you have load data in
a certain interval of time of a day or whole year. Apart from these, there are other
modes of power flow like harmonics, dynamics, fault study, etc. which are important
applications in distribution system studies.
The power solvers of distribution systems are generally formulated with radial
circuits in mind [27] which works very well in radial and weakly meshed circuits. But
in the case of OpenDSS, power solver was evolved from the program designed for
harmonic analysis making it quite different from the other distribution power flow
solvers [53]. OpenDSS uses primitive admittance matrix approach in building the
system model. A primitive admittance matrix Yprim is computed for each circuit
element in the model. These small matrices are then used to construct the system
admittance matrix Ysys. The procedure of generating Yprim is shown in fig 3-1and
equation 3.1.
Z is the 2x2 matrix describing the impedance characteristics of the coupled
impedance. As shown in fig 3-1, the two impedance system has 4 nodes or terminals
and equations to represent the current and voltage with respect to the zero-voltage
reference in each node can be written in the form of nodal admittance matrix. This
30
matrix relating the voltage and current is shown in equation 3.1. The primitive Yprim
is accumulated to make a system Y matrix(Ysys) and the new relationship is shown
in equation 3.2 where Ipc(V) represents the compensation current from all power
conversion elements in the circuit.
Figure 3-1: Building of Yprim for simple two phase Coupled Impedance [53]
I1
I2
I3
I4
=
Z−1 −Z−1
−Z−1 Z−1
V1
V2
V3
V4
(3.1)
IPC(V ) = Ysys ∗ V (3.2)
The currents injected from the PC elements in the circuit are the function of
voltage and represents the non-linear portion of currents from loads, generators,
storage and PV elements of the circuit. OpenDSS uses ”fixed-point iteration” method
to solve the non-linear equation as shown in equation 3.3.
Vn+1 = [Ysys]−1 ∗ Ipc(Vn) n = 0, 1, 2, ...until converged (3.3)
31
The iteration achieves a good convergence for distribution system when:
1. The initial guess of voltage is close to the final solution,
2. The series impedance of power delivery elements is less than equivalent shunt
impedance of the power conversion element. [53]
In the distribution systems, the first condition is easily achieved. During the
sequential time power flow solution, the voltage from the previous solution is taken
as start point which decreases the convergence time. To satisfy the second condition,
a limit of voltage is kept below which the power conversion models revert to the
linear model. Alternatively, OpenDSS also provides Newton power flow algorithm.
It is more robust than the fixed-point iteration for difficult circuits [54]. It takes
twice the calculations as compared to fixed point algorithm and is seldom required
in solving distribution systems. This algorithm is faster when performing long time
series simulation as Ysys remains constant during the simulation.
3.1.2 OpenDSS Architecture
The OpenDSS program is a script-driven simulation engine whose architecture is
represented in fig 3-2.
The engine has a main executive which can obtain the script from user written dy-
namic link libraries(DLL) or through the Microsoft Component Object Model(COM)
interface. The OpenDSS Executive is written in the Delphi programming language.
One of the reasons for using Delphi was to accommodate the COM interface with the
program easily [53].
The various components of the distribution system are divided into five object
class in OpenDSS:
• Power Delivery Elements: Line, Transformer, Capacitor, and Reactor
• Power Conversion Elements: Load, Generator, Voltage source, Current
source, Storage
32
Figure 3-2: OpenDSS Architecture [53]
• Controls: RegControl, CapControl, Relay, Reclose, Fuse
• Meters: Monitor, EnergyMeter, Sensor
• General: LineCode, LineGeometry, WireData, LoadShape, GrowthShape,
Spectrum, TCCcurve, XfmrCode
Each component has to be defined through the script. The typical script used in
our simulation are shown in Listing 3.1.
Listing 3.1: Script to define Various Distribution Components
New Line.feeder1 Bus1=31.1.2.3.4 Bus2=31_1.1.2.3.4 phases=4 Linecode=205
Length=1 Units=m !Script to define Line element
New Transformer.TD401346 windings=2 Buses=[65045 3865289.1.2.3.4]
Conns=[Delta Wye] kVs=[22 0.420] kVAs=[630 630] XHL=4.0 !Script to
define Transformer
New energymeter.m1 LINE.65043 1 !Script to define Energymeter
New Load.LOAD1 Phases=1 Bus1=1233.2.4 kV=0.23 kW=1 PF=0.95 daily=Shape_1
!Script to define Load
33
New Loadshape.Shape_1 npts=24 minterval=60
mult=(file=day_20_profile\shape_1.csv) useactual=true !Script to
define Loadshape
New Reactor.grnd1 phases=1 bus1=31.4 bus2=31.0 R=5 X=0.01 !Script for new
Reactor for Neutral Grounding
Figure 3-3: OpenDSS Interface with the master file
The main circuit model is made through sparse nodal admittance set of equations.
The Ysys is constructed from the nodal matrix of each circuit element. Circuit object
tracks the connection in the circuit with the data from individual circuit element.
34
The circuit elements on other hand are just aware of their connected bus.
The OpenDSS Interface needs a master file to link with the various circuit elements
file. Interface as shown in fig 3-3 represents a master file which links all circuit
elements. The OpenDSS Graphics Interface is suitable for basic operations and Power
Flow results but provides a very limited option for pre-processing of the data and
post-processing of the output results.
3.1.3 OpenDSS using COM Interface
The script-based interface of OpenDSS provides many applications regarding power
flow but has many limitations in applying new algorithms as well as post-processing
the output in the desired way. The OpenDSS engine is used for power flow solving
while other operations to change the network configuration and optimization of a
different variable can be performed in third party software by taking the output from
the OpenDSS as feedback. The code presented below allows MATLAB to run the
OpenDSS program through COM interface.
1 clear all;
2 clc;
3 DSSObj = actxserver(’OpenDSSEngine.DSS’);
4 if ~DSSObj.Start(0),
5 disp(’Unable to start the OpenDSS Engine’);
6 return
7 end
8 DSSText = DSSObj.Text; % Used for text interfacing from matlab to opendss
9 DSSCircuit = DSSObj.ActiveCircuit; % active circuit
10 DSSSolution=DSSCircuit.Solution;
11 % Write Path for master and associated file
12 DSSText.Command=’Compile (C:\User\Master.dss)’;
13 DSSText.Command=’batchedit load..* Vmin=0.8’; % Set Vmin lower so that
load model property will remain same
14 DSSTransformers=DSSCircuit.Transformers;
35
The COM interface can be launched through other popular programs like VBA
and Python. The DSS COM Interface has three basic children class:
• The Text provides user to launch DSS script command directly from the
MATLAB or another program.
• The Circuit provides access to the circuit elements that make up the circuit.
• The Solution gives access to the solution; making able to define solution
parameter, solve the circuit and view properties of the solution.
3.2 MATLAB R©
MATLAB software by Mathwork Inc. is one of the commonly used software in
scientific computing. Though MATLAB was designed for numerical computing its
scope is very wide and increasing allowing a great range of scope to the researchers4.
It is suited to the applications which deal with big matrices and vectors. MATLAB
has a wide range of embedded functions that can be used in numerical techniques. In
our application we used MATLAB for three purposes:
1. To read the GIS data and smart meter data in csv format and convert them
into the OpenDSS script,
2. To run the OpenDSS script using COM interface and extract the data from
monitors, and
3. To post-process the results obtained from the OpenDSS solver to display it in
a more presentable form.
MATLAB functions can be used to search for optimal solutions by use of their
embedded functions which is not available in OpenDSS. The other advantage of
programming through MATLAB is to observe the change in the circuit output with
the change in the circuit configuration in GIS. The MATLAB can also be used for
interactive programming to see the effect of different events in the network.4https://www.mathworks.com/products/matlab.html
36
Chapter 4
IEEE European Low Voltage Test
Feeder
In Europe, the distribution grids are quite different from the North American grid.
Most of the IEEE test system discussed in chapter 2 are designed keeping in mind
the North American system except the European LV test feeder. In the current
chapter, the basic feature of the IEEE European LV test feeder feeder is discussed.
The feeder is critically analyzed and some modifications are suggested to represent
the real European system.
4.1 Description of the test feeder
The test feeder is unique in nature compared to other test feeders as it has the
following main features:
• The test feeder is at voltage level 416 V which is a typical feature of European
distribution system.
• Load shapes are given in time series form for 24 hours every 1 minute giving
1440 load data points.
This feeder was introduced to include the diversity in the distribution feeders around
the world. The time series data is an important feature of this case as it plays vital role
37
in distribution research and planning. With the evolving demand side with storage
and PV in consumer premises, the static load flow for peak load is no more sufficient
for the distribution planning. The quasi-static data helps to study the influence of
various DERs in the distribution system. The time series data also provides the test
case to study the distribution system control using capacitor, regulators, and various
load balancing mechanism. The other important aspect of this test feeder is operating
frequency, which is 50 Hz unlike 60 Hz in other test cases.
Figure 4-1: One-line Diagram of European Low Voltage test feeder obtained from theOpenDSS test case
38
Voltage Source
The test feeder is connected to the medium voltage (MV) system through a
distribution transformer at the substation. The step-down transformer steps down
the voltage from 11 kV to 416 kV which is the operating voltage of the lateral feeders.
The one-line diagram of the LV-Test feeder is represented in fig 4-1. The MV system
of the feeder is 11 kV Voltage source whose impedance is specified by the short circuit
current, ISC3=3000 A and ISC1=5 A as shown in Master file 4.1.
Listing 4.1: OpenDSS Master File for IEEE LV test feeder
clear
Set DefaultBaseFrequency=50 ! for European system
New circuit.LVTest
Edit Vsource.Source BasekV=11 pu=1.05 ISC3=3000 ISC1=5 !Source Defination
Redirect LineCode.txt
Redirect LoadShapes.txt
batchedit loadshape..* useactual=no
Redirect Lines.txt
Redirect Transformers.txt
Redirect Loads.txt
Redirect Monitors_all.txt
New energymeter.m1 LINE.LINE1 1
Set voltagebases=[11 .416]
Calcvoltagebases
buscoords buscoords.txt !Linking Buscoordinate file
solve
Circuit Elements
The transformer at the substation is rated 0.8 MVA, rated voltage 11/0.416 kV
with delta/grounded-wye connection. The winding resistance and reactance are 0.4%
and 4% respectively.
39
The distribution lines are defined by the line code and length. Line code is
represented in form of sequence impedance and admittance. All lines are defined as
a three-phase line with multiple grounded neutral connection. The definition uses
the sequence component modeling where the effect of neutral resistance is considered
through reduced Carson’s equation [27]. Table 4.1 shows the sample line and line
code parameter required for simulation.
Table 4.1: Definition of Line and Line code in IEEE European LV test feeder
Line DefinitionName Bus1 Bus2 phases Linecode Length UnitsLINE15 15 16 3 2c .007 0.09654 m
Line Code DefinitionName nphases R1 X1 R0 X0 C1 C0 Units2c .007 3 3.97 0.099 3.97 0.099 0 0 Ω/km
The loads are constant PQ. The feeder has 55 single-phase loads. The load is
defined by kW and power factor. The load shape element defines the multiplying
factor for the load in every minute of a day. Each shape file has 1440 multipliers to
represent each minute of the day. Table 4.2 shows the definition of load where Model
Table 4.2: Definition of Load and Load-shape in IEEE European LV test feeder
Load DefinitionName Nphases Bus phases kV Model Connection kW PF YearlyLOAD20 1 349 1 0.23 1 wye 1 0.95 Shape 20
Load shape DefinitionName npts minterval File useactualShape 20 1440 1 Load profile 20.csv TRUE
1 refers to the constant PQ load. The phase in which load is connected is referred by
1, 2, or 3. The load-shape file links the load file with a .csv file which has 1440 values
for each minute of the day.
4.2 Limitation of the test feeder
The inclusion of long low voltage feeder with a large transformer and operating
frequency of 50 Hz makes the test feeder quite near to the European style grid. The
40
use of per minute time series data to represent data make it suitable for other studies
related to control and demand management. Despite the inclusion of maximum
characteristics of European style feeder few important points can still be modified
to make it similar to European style feeder.
Neutral Grounding
The test feeder assumes the line to be three-phase by reducing the neutral conductor
through Kron’s Reduction in Carson’s equation [40]. This might give a good result
for balanced load case and MV distribution system where neutral is grounded in the
span of few poles. But in the case of European Distribution System, LV lines are
longer, unbalanced and with isolated neutral. The neutral supplied to the consumer
is grounded only at the substation. The assumption in Kron’s reduction where the
neutral is earthed multiple times gives a wrong interpretation of consumer end voltage
[40]. The consumer earth ground and neutral are not mixed causing all neutral
current to return back to the substation through neutral wire. In this case, the
neutral voltage is above the assumed 0 V ground voltage. The neutral current flowing
through the neutral conductor causes the rise in the neutral voltage in the ratio of
neutral impedance. Further, the voltage received by the consumer will be phase to
neutral voltage instead of ground to earth voltage as assumed in the model. In the
real scenario, this voltage may rise up to the value which can sufficiently diminish the
voltage quality reaching to the consumer.
Per minute time series Load
The load presented in time series form is a very important feature of the test feeder.
But in the case of smart meter connected with PLC, it represents the unrealistic case.
Due to the bandwidth restriction in PLC [6] achieving 1-minute load from a smart
meter is almost impossible. In a trial with EDP HC Energia by the team, closest
reach was 20 minutes sample. The more efficient and error free data was obtained
when the sampling time for energy collection was kept for one hour. In another study
done by the team, it was observed that even in the sampling period of 10 second
41
conventional weighted least square (WLS) technique cannot be used for the state
estimation accurately. The load pattern can be changed to the one-hour basis for
making it similar to the real system or can be converted to per second time series
to check the possibility of real-time control and state estimation in the distribution
system.
Feeder type
The normal European network has transformers of 100 to 1000 kVA in size. Each
LV substation has multiple LV feeders originating from them. The LV network has
certain normal open (NO) and normal closed (NC) circuit breakers which allow the
possibility to change the network configuration. The IEEE European LV test feeder
misrepresent the actual European system as it has only one single feeder originating
from the LV substation with 0.8 MVA transformer. Further, there is no provision for
changing network configuration as it represents a single feeder in a single distribution
transformer.
Load Type
The European LV feeder consists of both single-phase and three-phase load. The
three-phase load can be balanced or unbalanced one. But the test feeder consists of
55 loads which are all single-phase. Further, the number of loads connected with 0.8
MVA transformer is quite unrealistic. Seeing the load pattern at the peak load, the
distribution transformer is loaded at 10% of its capacity which is very low compared
to the general norm of loading the distribution transformer around 70% during peak
load.
To address these limitations in European style test feeder, few changes are
proposed in section 4.3 and a new European style test system is proposed in chapter
6.
42
4.3 Modified IEEE European LV Test Feeder
In this section, two changes are proposed in the European LV test feeder. First, the
feeder is changed from three line model to four line model with TT neutral grounding.
In the second case, a new time series data for each second of one day is proposed.
4.3.1 Inclusion of Neutral Conductor TT Grounding
The IEEE LV test feeder uses the sequence impedance matrix instead of the primitive
matrix to reduce 4-wires model to 3-wires model.
Sequence Impedance Matrix Calculation
Figure 4-2 represents the model assumed in the European LV test feeder where the
primitive impedance matrix of a node is matrix Zprim given as:
Figure 4-2: Impedance modeling in IEEE European LV test feeder
[Zprim
]=
zaa zab zac zan
zba zbb zbc zbn
zca zcb zcc zcn
zna znb znc znn
(4.1)
The Zprim matrix is then reduced by Kron’s equation [40] assuming that the line
has multiple grounded neutral as shown in fig 4-2. The impedance matrix is converted
43
now to a 3x3 matrix where each element is given as equation 4.2 where zii represents
self-impedance and zij represents mutual impedance. The self and mutual impedance
used are obtained from modified Carson’s equation as shown in equations 4.3 and
4.4.
Zij = zij − zinznj/znn (4.2)
zii = ri + 0.0953 + j0.12134 ∗ [ln(1/GMRi) + 7.934] Ω/mi (4.3)
zij = 0.0953 + j0.12134 ∗ [ln(1/Dij) + 7.934] Ω/mi (4.4)
The 3x3 matrix obtained after Kron’s reduction is the reduced impedance matrix
Zabc given as equation 4.5 after partition between the row and column of phase c
and row and column of neutral.
[Zabc
]=
[zabc
]−[zan
] [znn
]−1 [zna
](4.5)
The sequence impedance in European LV test feeder is represented by zero (Z0)
and positive (Z1) sequence impedance as shown in table 4.1. The diagonal terms
of the sequence impedance matrix are zero (Z0), positive (Z1), and negative (Z2)
sequence impedance. Z1 and Z2 are considered equal in the case of the radial
distribution network. Further, to model the system in the base of sequence impedance
the following assumptions are made:
• All self-impedance should be equal.
• All phase to phase mutual impedance should be equal.
• All phase to neutral mutual-impedance should be equal [40].
So, the simplified Zabc can be written as 4.6 where Zs is reduced self-impedance and
Zm reduced mutual impedance.
[Zabc
]=
Zs Zm Zm
Zm Zs Zm
Zm Zm Zs
(4.6)
44
The sequence impedance matrix Z012 can be represented in form of reduced phase
impedance matrix Zabc as equation 4.7.
[Z012
]=
[A]−1 [
Zabc
] [A]
where,[A]
=
1 1 1
1 a2 a
1 a a2
(4.7)
On further calculations of equation 4.7 sequence impedance matrix is obtained
as
[Z012
]=
Zs + 2 ∗ Zm 0 0
0 Zs − Zm 0
0 0 Zs − Zm
(4.8)
The sequence impedance will now become:
Z0 = Zs + 2 ∗ Zm (4.9)
Z1 = Z2 = Zs − Zm (4.10)
Primitive Impedance Matrix Calculation from Z012
For making the modified IEEE LV test feeder have the same impedance as the base
case, reverse equations were formed to derive the primitive matrix without ground
return path since all neutral current is assumed to return from the neutral conductor.
Z1 = zii − zij = ri + jxi (4.11)
In equation 4.11, it is observed that Z1 represents the primitive self-impedance
when the return path of ground, i.e., earth resistance is not considered. This
neutral impedance can be calculated based on the neutral sizing. But there is no
documentation about conductor sizes, so it is assumed that neutral impedance is
equal to the phase impedance. Further, any mutual impedance is neglected because
in simplified Carson formula earth impedance forms the mutual impedance.
45
Figure 4-3: Modified Impedance modeling as per real European style TT grounding
Modification in OpenDSS code
The figure 4-3 represents the actual European style LV feeder where neutral is
grounded at the substation and it is isolated from any other ground. The return
path of consumer neutral current is provided by a neutral conductor. OpenDSS has
a special feature which allows simulating this system as a four-phase system with few
modifications in codes.
The line code in OpenDSS is now converted for 4-phase with the primitive matrix
excluding any mutual impedance and dirt resistance.
New LineCode.2c_.007 nphases=4 baseFreq=50 units=km
~ rmatrix = [3.97 | 0 3.97 | 0 0 3.97 | 0 0 0 3.97]
~ xmatrix = [ 0.099| 0 0.099 | 0 0 0.099| 0 0 0 0.099]
The line is converted from default three-phase system to four-phase considering
the neutral. OpenDSS takes Bus1=10.1.2.3.0 when we write Bus1=10 and assumes
that the neutral is at the same potential as the ground at 0 V. But to simulate
isolated neutral Bus1=10.1.2.3.4 expression should be used where .1.2.3.4 represent
three phases and neutral.
New Line.LINE1 Bus1=1.1.2.3.4 Bus2=2.1.2.3.4 phases=4 Linecode=4c_70
Length=1.098 Units=m
New Line.LINE2 Bus1=2.1.2.3.4 Bus2=3.1.2.3.4 phases=4 Linecode=4c_70
Length=0.11511 Units=m
46
The transformer is modified to represent the delta-wye connection with star
point available for connection in wye side (Low Voltage side). Delta side is connected
with three phase MV line without neutral.
New Transformer.TR1 Buses=[SourceBus 1.1.2.3.4] Conns=[Delta Wye] kVs=[11
0.416] kVAs=[800 800] XHL=4 sub=y
To represent the transformer neutral grounding, a reactor element is defined
connected to the fourth phase (star point as shown in figure 4-3) with the ground.
It is assumed that the ground resistor in the distribution system remains around 5Ω
based on the feedback of distribution utility in Spain.
New Reactor.ground phase=1 bus1=1.4 bus2=1.0 R=5 X=0.01
The load element is defined as 1 phase connected with respective phase and neutral
(.4) instead of the ground as seen below
New Load.LOAD1 Phases=1 Bus1=34.1.4 kV=0.23 kW=1 PF=0.95 Yearly=Shape_1
New Load.LOAD2 Phases=1 Bus1=47.2.4 kV=0.23 kW=1 PF=0.95 Yearly=Shape_2
Observations after Modification
The voltage pattern was observed after the modification of the European LV test
feeder to accommodate the neutral wire.
It can be seen from fig 4-4 that when the neutral conductor is considered, the
voltage obtained by the consumer is lower than the original model where the return
path is considered as ground. The voltage difference is high at the point where the
neutral voltage is increasing. The neutral voltage is produced due to the voltage
drop in the neutral conductor, which in the original case was neglected assuming the
return path to be earth and earth has less impedance than the neutral wire being
used in the distribution system. In another observation represented in fig 4-5, the
neutral voltage is observed to rise up to 10 Volts. The cases might deteriorate if the
load increases and the unbalances become extreme. In the load data provided in the
47
Figure 4-4: Phase B Voltage(Ph-ground) in all nodes in (a) Original and (b) ModifiedEuropean LV test Feeder
48
time (min)
0 500 1000 1500
Volts
0
2
4
6
8
10Neutral Voltage observed in all nodes after modification in European IEEE LV test feeder
time (min)
0 500 1000 1500
Pow
er(
kW
)
-80
-60
-40
-20
0Total Power Supplied by Transformer
Figure 4-5: (a) Neutral Voltage in all nodes for a day with European style TTgrounding in modified IEEE European LV test Feeder (b) Total Power Supplied bythe Distribution Transformer
test case, the transformer is very sparsely loaded, i.e., the peak load is less than 70
KW that too for very less time (10% Peak Loading). It is also seen that the peak
in neutral voltage is observed when the load is at the peak. The problem of neutral
voltage will increase in the feeder and cannot be neglected as done in IEEE European
LV Test Feeder.
Figure 4-6 shows the difference between modified and original test feeder in terms
of the node current in a line. As observed in the top figure representing the four-wires
model, the fourth wire represents the neutral wire and the current is the vector sum
of all phase currents. The current returned from the ground is 0 A. In bottom figure,
representing the 3-wires model, the same return current is seen but passing through
the ground. It shows that the top figure represents the European style feeder in a
more realistic way than the original test feeder. The assumption that the potential
of neutral to be equal to ground potential i.e., 0 V at every node in the original
European test feeder can be avoided if the modified model is adopted. This will help
to understand the real unbalanced simulations in the European style LV distribution
networks. The neutral voltage profile in figure 4-7 along the node shows that towards
the end of very long feeder the neutral voltage is more prominent. The test feeder
49
Figure 4-6: Comparison of Current in different node for Line 170 between Modifiedand Original European LV Test feeder
Figure 4-7: Neutral Voltage profile along the feeder
50
presented is just 500 m long. But in many cases, feeder length is longer than that
creating a greater neutral voltage. Neutral voltage study thus provides an important
tool for distribution system planner and must be included in the test feeder.
4.3.2 Per Second Time-series Load Data
As part of a research project at Vienna University of Technology a data set of per-
second resolution of electric power profile of 30 households in Austria was recorded.
The Data was generated in the research project “ADRES-CONCEPT” (EZ-IF:
Development of concepts for ADRES – Autonomous Decentralized Regenerative
Energy Systems, project no. 815 674) [55]. Using least mean square error in per
day consumption, per-second profile with the lowest error was assigned to each IEEE
load profile with energy factored by the total energy of the IEEE load. The new load
profile thus will have the same energy consumption at the end of the day but will have
per second resolutions with 86400 data points. Change in the OpenDSS code was
done to accommodate the per second load data by changing the loadshape element.
New Loadshape.Shape_1 npts=86400 sinterval=1
mult=(file=IEEE_modified_100profiles\New_Match_IEEE_P_1.txt)
useactual=true
New Loadshape.Shape_2 npts=86400 sinterval=1
mult=(file=IEEE_modified_100profiles\New_Match_IEEE_P_2.txt)
useactual=true
The aim was to provide researchers with load data of lower sampling period and also
to see the impact in overall balancing while considering the per-second resolution of
the load.
Results of Modification
The neutral voltage profile observed in the modified case was less than the original
case because the load becomes more balanced in taking a lower resolution.
The figure 4-8 represents the high-resolution phase current and the neutral voltage
51
Figure 4-8: Phase Current and Neutral Voltage at Bus 558 after Load modificationto Per-second load
at node 558. The lower neutral voltage shows that the load is more balanced on
lowering the resolution but further researches are required to verify it. The modified
load pattern can be used to study the distribution state estimation.
Still, after the modification, many limitations of the IEEE European test feeder
as discussed in section 4.2 could not be addressed. In the next chapter, a method is
proposed to convert the real LV network of a city in Spain to the real European test
network to address all the limitations.
52
Chapter 5
Integration of Raw data to
OpenDSS Code
One of the purposes of this master thesis is to deliver EDP HC Energia an OpenDSS
model of distribution network taking in account the network data, load position
and smart-meter readings. The model should be able to accommodate any changes
and configuration of the network structure. It should also provide the good visual
representation of the network status for the distribution planning as well as possibility
to add storage and renewable elements in future. Based on the experience of
converting the IEEE European LV test feeder to 4-wires model as discussed in chapter
4, the new model is directly simulated in 4-wires format. Further, the way to convert
the GIS data directly into indexed form was also explored, so that the data will be
available to researchers without the breach of any privacy. In the following sections,
the network data obtained from the GIS application explained along with the strategy
for their conversion in an OpenDSS format with proper re-annotations.
53
5.1 Description of Raw Data
The raw file for the network was in xlsx format with name Red NAME DATE.xlsx 1.
The description of the load was extracted with name Inventario Contadores NAME.xlsx 2.
The smart meter energy readings were extracted in Curva Carga Contadores NAME.csv 3.
Due to the limitation of the excel file, the files for smart meter energy reading were in
multiple files to accommodate all meters and their consumption record. The phase of
the loads was determined by PLC communication and field verification and included
in phase meters.xlsx. The first three files were extracted from Bentley GIS tool4 being
used by EDP HC Energia using Oracle database handling tool.
Figure 5-1: Description of Raw file used to Create OpenDSS model of largeDistribution Network
5.1.1 Network Data (red)
The Network data consist of the details of all distribution elements component in
LV side. The Spanish distribution system generally have manual tap-changers in
1Red is the Spanish word for ’network’2Inventario Contadores is Spanish words for ’Inventory Counter’3Curva Carga Contadores is Spanish words for ’Load Curve from Smart Meter’4https://www.bentley.com/en/products/brands/openutilities
54
11/0.416 kV substations but they are operated very seldom and is neglected in this
model. The line component and fuses used to change the network configurations are
the important components of the a distribution system which defines the topology.
Figure 5-1 shows the various sub-file included in Network file. There are six sub-files
in Network file which have been analyzed and converted to OpenDSS Model.
Transformer (CT-Trafo)
Transformer data consist of the details of distribution transformer. As shown in
figure 5-2 transformer file consists of 11 columns. Of them, column G represents the
unique code to the transformer. In the simulation, it is considered as the name of the
transformer. Column F (Mslink Celda) represents the bus 1 of the transformer, i.e.,
Figure 5-2: File from GIS representing Transformer in Network
MV side and column K (Nudo Celda Trafo) represents the bus 2 of the transformer
i.e., LV side. The column J represents the capacity of the transformer (Potencia
Trafo) in kVA. The connection for all transformer is taken as delta-wye and the %
impedance is 4%. The column D and E represent the X and Y coordinates of the
substation. The MV and LV side of the transformer are assumed to be in the same
coordinates.
LV Circuit Breaker at Substation (Linea BT)
The circuit breaker data at substation consists of the details of the feeder originating
from each substation and their status as shown in figure 5-3. The entry in column E
(NO) represents the origin bus which links with bus 2 of the transformer. It signifies
the transformer in which the breaker is connected. The column B (Clave Linea)
represents the feeder number for feeders in each transformer. The column A (Mslink)
is the unique ID for each feeder. The column I (EST Normal) and column J (Est
55
Figure 5-3: File Representing the LV circuit Breaker at the origin of feeder
Operac) represent the normal condition and present condition of the breaker. The
value C refers the breaker is closed5 and A refers the breaker is open6. The column
G and H represent the X and Y coordinates of the bus2 of the breaker. The length
of the line connecting the LV side of terminal and breaker by distance formula in the
coordinates from bus 2 of transformer represented bt Trafo x and Trafo y (Column D
and E of CT Trafo) and Bus 2 of the Feeder Breaker represented by Lin x and Lin y
(Column G and H of Linea BT ), the length of feeder is given by:
Distance(d) =√
(Trafo x− Lin x)2 + (Trafo y − Lin y)2 (5.1)
Line Segments (Segmento BT)
The line segments file has details of segments of each feeder defined in LV circuit
breaker(Linea BT ) as shown in figure 5-4. The column A (Mslink) is the unique Id
Figure 5-4: File Representing the Line segments of each feeder
for each segment. Column B represents the length of the segment in meter. Column
D (Mslink Linea) is the unique code for each feeder. Column F (Nudo Origen) is the
Bus 1 of the segment while Column G (Nudo Destino) is the Bus 2 of the segment.
Column H represents the type of the cable (Tipo Cable) of that segment.
5C refers cerrado which means Closed6A stands for abierto in Spanish which means Open
56
Line Segment Coordinates (Coordenadas Segemntos)
The line segment is the line between two points where there are same cable type
and no branching. The line segment parameter is enough to calculate the nodal
power flow. But if we want to represent the line segment in the map, it requires the
coordinates. The ’line segment coordinates’ is the file which has data of all the line
segments as shown in figure 5-5. Each line segment can have two or more coordinates
Figure 5-5: File Representing the Coordinates of each Line segments of each feeder
based on the type of line. If it is a straight line then two coordinates will be enough
to represent the line segment but if it has curves or change in direction, it needs
more coordinates to represent it in the map. The column A (Mslink) is the unique
code assigned for each segment. The part of the segment is represented by column F
(Orden Vertice) whose value starts from 0 and ends with the number of coordinates
required to represent the line segment. The column G and H represent X and Y
coordinates of the part of the segments that form the line segment. This file can
be neglected if we do not want the visual representation of the network but is an
important file in our case as we want a visual representation of data.
Load Connection Point (Acometidas)
The load connection file as shown in figure 5-6 provides the link between the loads
and the network. Column B (Clave BDI ) of this file is the unique load connection
point Id which connects the load in smart meter record file. The bus in the network
in which this load connection point is connected is given in column C (Nudo Origen).
57
Figure 5-6: File Representing the Load connection points in the network
Circuit Breaker in Network (Fusible)
This file plays an important role to model the change in network connectivity. The
LT network of a city consists of many breakers which are normally open or closed and
can be used to change the network configurations. As shown in figure 5-7, column
Figure 5-7: File Representing the position and status of Circuit Breaker in network
D (Nudo origen) and F (Nudo Extremo) represent the bus 1 and bus 2 of the circuit
breaker. The column F (Est normal) is the normal operating condition of the circuit
breaker while column G (Est operation) represents the current satus of the circuit
breaker. As in circuit breaker at the substation, A refers open condition (abierto)
and C refers to the closed condition (cerrado) of the circuit breaker.
5.1.2 Smart Meter Record File (Inventario Contadores)
The smart meter record file consists of all the smart meter in the network of interest.
The file has all the information of the load ranging from the transformer connected
to, full address of the consumer and many more. But for the load flow simulation
and visual representation, the only point of interest is the connection point of the
smart meter in the network. As shown in figure 5-8, column A of the data has the
unique ID of each smart meter (referencia) and column R has the unique Id of load
58
Figure 5-8: File Representing the Smart meter in the network
connection point (acometidas) defined already in network file.
5.1.3 Load Curve File (Curva Carga Contadores)
Load curve file is required to define the time series load data for each load as shown
in figure 5-9. It is the data obtained from the smart meter every hour. In our model
Figure 5-9: File Representing the Load Curve
we are interested in the active power for the load which is represented in column G
(Active E ). Column D (fecha) represents the time to which the active power belongs
to while the column C (referencia) is the unique Id for the load. Each load will have
24 entries for a day and can be more if the days of interest are more. In few cases,
due to missed communication, some of the loads have less than 24 entries for a day.
5.1.4 Phase Data
This is the file with phases of the all connected load in the network. The phase
data was obtained by the field verification of the results obtained from PLC
communication. As shown in figure 5-10, the file consists of unique Id for load
59
Figure 5-10: File Representing the Phase in which Load is connected
(referencia) in column A and phase in column B. The R, S and T indicate phase A,
B, and C respectively for single-phase load, while RST represents 3 phase loads.
5.2 Indexing of Circuit Element Data
The aim of this thesis is to generate the test network through the real network data.
But as mentioned in [1] this requires conversion of data to remove private data. Also,
the network elements should be re-annotated to make data analysis and representation
easier. This section deals about the re-annotation process of all the data.
Load Indexing
The smart meter file is first sorted by CT number and then alphabetically and the
load is renamed with integers starting from 1. Since all load names are unique it does
not have any issue. The bus connected is linked with the load during indexing.
Line Indexing
The overall line is divided into three types: MV line, LV busbar and LV line. The MV
lines are the lines connecting the source with the transformer. The LV busbars are
the lines representing the segment from the LV side of the transformer to the circuit
breaker.
Further, the circuit breaker in substations and between the LV network are also
assumed as lines and are indexed differently. All lines are indexed in the format
[xxxxx00], where xxxx represents the type of line and [00 00] represents the unique
60
number for each line of the same type.
Table 5.1: Description of Codes used in line Indexing
Type of line String[xxxx] number[00 00]
Medium Voltage mv 1,2,3...LV Feeder from Transformer feeder 1,2,3...LV breaker in substation cktbk 1,2,3..(for closed ckt only)LV line [blank] 1 0,1 1,1, 2 0,2 1,2 2,2 3,...LV Breaker in mid-network fuse 1,2,3,.. (both for closed and open)
As shown in table 5.1, [mv00] is used to signify the MV lines, [feeder00] is used to
represent the feeder line from Transformer to LV Breaker in the substation, [cktbk00]
to represent the circuit breaker in the substation and [fuse00] to represent circuit
breaker in the middle of LV network. Remaining lines are simply represented by an
integer [00] if they represent the monitored segment of line and by [00 00] if they
are non-monitored segment, where the first integer represents the line segment and
second the sub-segment of that line segment.
The indexing of the line was done to prevent the coordinates of the line segment
from generating more monitored buses. As discussed in the network data, each line
segment may require more than two coordinates to represent correctly in the map.
But as discussed in [38], it is not necessary to monitor all those small segments. There
is a need to define a separate line to visually represent the feeder as it is but different
naming method is adopted to make the monitoring easy.
Figure 5-11 shows a simple example for LV line indexing scheme so that there is
no need of further reduction of the circuit as researchers did for IEEE LV test feeder
in [38]. It is observed that the line segment between point a and b is a line segment
which is straight and do not need any segment, so the line code is 3. At point b, two
line segments originate 4 and 5. For line 4 it is seen that representing it needs more
coordinates (0,1,2,...,8) which means that it needs 8 lines to define as it is in the map
but since there is no branching or any load in between, the current will be same for
all of these segments. So instead of defining them all with an integer, only the last
sub-segment is named as line 4 while the sub-segments are indexed as 4 0, 4 1, etc.
61
Figure 5-11: Simple Line Topology for Line Indexing
The advantage of this system is that while monitoring, one measuring tool (monitor
in OpenDSS) would be enough to represent all these lines sub-segments.
Bus Indexing
Figure 5-12: Simple Line Topology for Bus Indexing
The strategy used in bus indexing was to name the bus 2 of each line same as the
line. The advantage of doing so is that it makes monitoring of the line easier and less
62
confusing. Further in OpenDSS when you monitor the line the observed parameters
are of Bus 2. So, making them with same name makes possible, monitoring the line
and bus with less complexity.
As shown in figure 5-12, the line code for the segment ’ab’ is 3. So, the bus 2
of the segment, i.e., point b is indexed as 3. The segment arising upward from point
b has multiple segments to define coordinate and are indexed as 3 1,3 2,...,3 8. The
end of the segment is the point of our interest so will be indexed by integer 4, which
is also the name of the line joining Bus 3 8 and 4. The bus number is first assigned
to the all HV side of the transformer, then to the LV side. The busbar of LV side
is then assigned the name in the format [00 00]. As seen in the figure 5-12, if the
system has only one transformer, HV side bus will be named 1, LV side as 2 and
Bus 1 of circuit breaker as 2 1,2 2,....,2 8. The point to note here is we have omitted
the LV busbar where the circuit breaker is open but retained their name to maintain
uniformity. So, in the above case, bus 2 2 and 2 3 does not exist but will come in the
role if the respective breaker is closed. Table 5.2 shows the number reserved for the
different type of bus if the system has ’nt’ numbers of transformers and ’ns’ numbers
of line segments.
Table 5.2: Description of Codes used in Bus Indexing
Type of Bus Number Assigned
Medium Voltage Side of Transformer 1,2,3..,nt (nt=transformer number)Low Voltage Side of Transformer nt+1,nt+2,...,2*nt)Bus 2 of Line segment (2*nt+1),(2*nt+2)...,(2*nt+ns)Starting Bus after Breaker (2*nt+ns)+1,(2*nt+ns)+2,....Substation Breaker Transformer side(Busbar) (nt+1) 1,(nt+1) 2,...Sub-segments of the line except first and last bus (2*nt+1) 1, (2*nt+1) 2,...
5.3 Complexities in Data
The indexing method explained in section 5.2 solves the problem of hiding the private
data as well as avoiding the monitoring of the small segments which were taken just
for the visual representation of line segment. It also makes the model cleaner with
63
shorter and easier name than the original data. But the data has a certain problem
which makes it difficult to model directly.
Bus Indexing Complexities
The major problem was the improper definition of bus 1 and bus 2 in line segments
and circuit breaker at mid-circuit. It was observed that in line segment (Segmento
BT ) file, the column F and G representing origin and destination did not have the
proper orientation, i.e., either of them could be origin or destination. During indexing,
the segments are named on the basis of the destination bus (Bus 2). So, knowing the
right destination bus is very important for building a correct network.
Figure 5-13: Simple Line model to Explain the Data Complexity
To explain this problem, an imaginary case is considered in figure 5-13 which is
a network with 10 lines and 11 bus. The data in table 5.3 represents the raw data of
the line segment. If they are used for indexing without any modification, there will
be duplication in the line name as the all three lines a, b and c has now name ’line
2’. Further, the monitor element of the line b and c will give the value in opposite
sign compared to line a as the direction of line b and c is opposite to the direction of
current flow.
64
Table 5.3: Example of Line Segment Data in Raw form
line segment Bus 1 Bus 2
b 3 2a 1 2c 10 2d 4 3e 3 11
A simple solution of this problem can be adding some alphabet after 2 in line b
and c if they already appear in other lines but it will not solve the actual problem for
which indexing was considered in the first hand. The other solution can be naming
the line by Bus 1 name in case Bus 2 name is already taken by other lines. This
solution solved a problem for only if the starting lines orientation is correct. But in
raw data, the line segments were not ordered as in the example above. In such case,
the orientation of the whole feeder becomes opposite. Further, in the radial network
in the distribution system, a bus can be origin point of many line segment but can be
destination point to only one line segment. So, if the sub-segments of line and line
segments are designated by destination bus it is always unique. But in this solution,
there are chances we have to rename using alphabets, which once again will not solve
the problem. The same problem existed in the case of the circuit breaker (fusible)
definition. Due to the wrong origin and destination bus position, in a few cases, the
segments after the circuit breaker were isolated.
Further, there were a few lines which were not connected to the source and a
few loads which did not have any connection points. These are the floating points
in the GIS network due to mistakes in GIS indexing. These cases should be filtered
out and checked with the utility regarding the physical existence of those feeders and
update the same in GIS. In the next section, the solutions to these problems have
been discussed.
65
5.4 Algorithm to re-annotate the Bus, Line, and
Load
As discussed in the previous section arranging the origin and destination bus is
important to model the network correctly. To achieve this in the first stage, the
Bus 2 of the circuit breaker in the substation is used to find the origin point of every
feeder using the algorithm shown in figure 5-14. In the second part, the origin and
destination bus for following line segment is checked and rearranged if required along
with the coordinates data in file Coordenadas Segmentos.
Both algorithms are based on the radial nature of the LV network that every bus
can be destination bus to only one line but can be the origin point for many lines.
In the first part, the destination buses of all circuit breakers in the substation which
are closed are taken as a starting point to find the starting point of the line segment.
If the destination bus of the breaker is found in origin bus of the segment then its
destination bus is stored in a [segcon] vector with the segment code (It is the position
of a line segment in Segmento BT file). But if it is found in destination bus column,
then the exchange of bus is done for that segment and then destination bus is updated
in [segcon] vector. At the end of the loop, a vector [segcon] with destination bus of
all first feeders in the substation is created.
Algorithm for Line Segment Segment and Circuit Breaker in the Network
Once the first segment of the feeder with their origin and destination feeder is known,
an attempt is done in each loop to find the consecutive feeder. Each time if some
destination bus is found in origin bus position of another segment, the [segcon] vector
is updated with the destination bus of the new segment with the segment code. On
the other hand, if the destination bus of any segment is already in [segcon] vector,
the exchange of bus is done along with rearranging of the coordinates of the segment.
In this process, only the segments which are not already in the [segcon] vector are
checked. At the end of every loop, check for connection with the fuse element is done.
A vector [closed fuse] is created with both origin and destination bus of all closed
66
start
i<=size(segment)
yes
no
Create Vector [closed_start] It includes Bus2 of all closed
breaker inSubstation
i=1 segcon=[ ]
Segment(i)_Bus1 is in [closed_start]
yes Add Segment(i)_Bus2
in [segcon]
i=i+1
Segment(i)_Bus2 is in [closed_start]
Interchange Bus1 and Bus 2 in Segment(i)
rearrange the coordinate file
yes
Vector [segcon]is vector of all
first linecoordinates
no
End
Segment(i)_Bus1 and Bus2 are unique is in [closed_start]
Only Segment(i)_Bus2 is Unique
Only Segment(i)_Bus1 is Unique
Interchange Bus1 and Bus 2 in Segment(i)
rearrange the coordinate file
yes
yes
yes
no
no
no
Figure 5-14: Flowchart to link the circuit breaker with first point of line segment
circuit breaker in the network. When either of the buses is found in [segcon], the
other bus is updated in [segcon] with the segment code as 0. The loop continues until
67
start
i<=size(segment)
yes
no
Load Vector [segcon]
from previous loop Create vector
[closed_fuse](mx2) with Bus 1 and 2 of
closed CB
define len_diff=1
Segment is already in [segcon]
yes
Add Segment(i)_Bus2
in [segcon] segmentcode=(i)
i=i+1
Only Segment(i)_Bus2
is in [segcon]
Interchange Bus1 and Bus 2 in Segment(i)
& rearrange the coordinate file
yesOnly
Segment(i)_Bus1 is in [segcon]
no
no
Update [segcon] withBus 2 of fuse_closed(j)
segmentcode=0
no
Check if any Bus in[segcon] linked with
Circuit Breaker j=1
len_diff>0
len1= Size of [segcon]
i=1
yes
yes
j<= size[fuse_closed]
yes
Only Bus1 of fuse_closed(j)
in [segcon]
no
Only Bus2 of fuse_closed(j)
in [segcon]
j=j+1
Update [segcon] withBus 1 of fuse_closed(j)
segmentcode=0
no
yes
yes
no
len_diff=len(segcon)-len1
no The [Segment] fileis now ready for
Indexing end
Figure 5-15: Algorithm for arranging Bus 1 and Bus 2 of Line Segment correctly
68
the size of [segcon] stops increasing.
Table 5.4: Working of Algorithm to check the position of Bus
Original Data Loop 1 Loop 2
Seg Bus 1 Bus 2 Bus 1 Bus 2 [Segcon] Bus 1 Bus 2 [Segcon]
b 3 2 2 3 [2,3]
a 1 2 [2,3]
c 10 2 2 10 [2,3,10]
d 4 3 3 4 [2,3,10,4]
e 3 11 3 11 [2,3,10,4,11]
i 7 6 [2,3,10,4,11] 6 7 [2,3,10,4,11,6,7]
g 6 4 4 6 [2,3,10,4,11,6]
The algorithm is used to explain the arranging of the data given in Table 5.3
assuming the [segcon] has already the data of segment a. So, the initial value of
[segcon] will be [’2’]. The algorithm will check in sequence to see if any Bus 1 or
Bus 2 value is already in [segcon] vector. When it reaches first entry, which has
Bus1=’3’ and Bus2=’2’. Since ’2’ is already in [segcon], the bus name of Segment
’b’ is swapped with Bus1=’2’ and Bus2=’3’. [segcon] will be updated with newly
discovered destination bus ’3’, i.e., [segcon]=[’2’,’3’]. The second segment is neglected
as it is already in [segcon]. Similar things happen in the third and fourth segment,
where Bus 1 and Bus 2 are swapped with an update in [segcon]. In the fifth segment,
it is seen that the Bus1 is in [segcon] vector. So in this case, the Bus1 and Bus2 are
not swapped but keep as it is while [segcon] vector is updated with Destination Bus
’11’ of this segment. At the sixth segment, neither Bus 1 nor Bus 2 is in [segcon]
so this segment is skipped. In the seventh segment, again Bus 2 is in [segcon] so
swapping of Bus 1 and Bus 2 is done and Bus’6’ is updated in [segcon]. In the second
loop, all the segments except ’i’ are neglected as they have been all updated. Now, in
the sixth segment, Bus ’6’ is in [segcon] so swapping of Bus1 and Bus2 is done with
an update in [segcon].
The other vector in used is [fuse closed], which has data of all Bus1 and Bus2
69
of the closed circuit breaker. At end of every loop, the elements of [fuse closed] are
compared with the destination bus stored in [segcon]to find any linking bus. This
process incorporates the segment after the circuit-breaker to arrange them with the
proper origin and destination feeder to re-annotate the bus name and line name
correctly. The whole loop continues until the size of the vector [segcon] is increasing.
Algorithm for Load Indexing
As discussed in the previous section, load data had few problems which need to be
sorted out for the proper running of simulations. The following method was used to
rectify those issues and create a working model.
• Phase missing: The phase data obtained from the field was not complete
especially due to the changing loads or meter as well as lack of a proper database
on utility. Phase identification of the load in the distribution network itself is
a huge topic with many researchers working on it [56, 57]. It is an important
aspect of real modeling but was not included in the scope of the thesis. For
approximately 5% of cases where load phase data was not available, a random
phase was given.
• Floating Load: Few cases of loads and line segment which were freely floating
without connections to any source was found. This cases requires field survey
and updating in the GIS model. Since there were only less than 5 cases in a
network with 8087 loads and 12,000 nodes, these were manually removed based
on the Isolated node report generated by OpenDSS.
• Non-Smart meter: There were still a few meters which were not smart. Those
meters had no load data for hourly consumption. For now, the consumption of
such meter is considered to be 0 kWh for simplicity. But it is expected that all
the meters will be the smart meter in near future.
70
5.5 Program Structure for Data Extraction
A modular script in MATLAB was created to convert the GIS data file into the
indexed OpenDSS file for the whole network. The MATLAB script was divided into
two basic parts; first to generate the constant part of network, i.e., transformer,
distribution lines, breaker and load. The second part was to extract more variable
parameters of load consumption from smart meter record.
5.5.1 Network Data
The network data file can be extracted and converted into OpenDSS format by
shortcut.m which contains the files to extract the data from GIS file and create
indexed OpenDSS network file as shown following:
1 %shortcut.m %Set of File to Generate OpenDSS files
2 [transformer, Line, lnraw, Segment, Segraw, buscord, loadbus,txID, fuse,
fuseraw, load_loc, load_loc_raw, phaseraw]=filename;
3 linecode;%generates line code for existing type of feeder
4 bus_cord_modified; %Indexing of Bus number before rearrangement
5 Seg_crazy; %arranges BUS 1 and BUS 2 in proper order
6 bus_cord_modified; %Indexing of Bus number after rearrangement
7 line_indexed; %OpenDSS file generator for LV_line
8 bus_cord_indexed; %OpenDSS file Generator for Bus Coordinate
9 source_ind; %OpenDSS file Generator for MV line from source to transformer
10 fuse_indexed; %OpenDSS file Generator for Circuit Breaker in LV network
11 reactor_neutral; %OpenDSS file generator for Transformer
12 transformer_indexed; %OpenDSS file Generator for transformer
13 energymeter; %OpenDSS file Generator for Energy Meter
14 load_indexed; % OpenDSS File Generator for Load
The different functions used in the shortcut.m are explained below:
1. filename.m: This function is meant for extracting data from the excel file
in MATLAB. The function has three variables, ’file master’ (Network file),
71
’file load’ (Smart Meter File) and ’file phase’ (Load Phase File) to give the
file name of raw file on 5-1 manually. These files pointers are used to access
those excel files and their sub-files.
2. linecode.m: This function gives the linecode.txt file which has linecode of
all type of line described in Line Segments (Segmento BT ). We attempted to
make a universal file by the inclusion of all possible distribution cable type in
Spain. The line code is generated in 4 × 4 format to include the neutral point
and all mutual impedance are neglected.
3. Seg crazy.m: This file is meant for re-arranging the wrongly placed bus. It
reshuffles the origin and destination buses in line segments where it is wrong
based on the algorithm discussed in figure 5-14 and 5-15.
4. bus cord modified.m: This .m file is used for indexing the bus name. The
file takes all the bus name from line segments and removes the duplicates. Then
indexing of the segments is done as mentioned in table 5.1. This code is run
twice, before the arrangement and after the arrangement. The second run is for
making the line index and index bus 2 of that line same as discussed in section
5.2 to make easier in monitoring.
5. line indexed.m:This file is meant for generating LV line details including
the substation breaker and LV feeder. This file generates all the sub-segment
of the line segment to give its length, line codes and connecting bus as per
the indexing pattern discussed in section 5.2. Further, it also incorporates
the 4 lines, including the isolated neutral which allow the return path for the
unbalanced current. At the same time, it also generates monitor file for each
feeder at the substation and all bus in the LV network.
6. bus cord indexed.m: This file generates the bus coordinates of all the LV
buses based on the coordinates in the raw data file and the name as per the
new indexing.
72
7. source ind.m:This file is to generate the MV lines in our circuit. In this
simulation, MV lines of 5 m each from a common source to all the transformer
are generated. The effective resistance of MV lines is assumed to be very low
i.e., 2.5 × e−6Ω in this simulation as we are interested in LV part only. At the
same time, it also generates the monitors for MV lines.
8. fuse indexed.m: This file is to create the file for all circuit breaker in the
middle of the LV network. The data is extracted from fusible file and indexed
according to the indexing rule. The effective resistance of the circuit breaker is
assumed to be 25e−6Ω.
9. reactor neutral.m: This file generates circuit objects used in the neutral
grounding. It is assumed that the transformer neutral is grounded with
the resistance of 5 Ω and very nominal impedance. This incorporates our
assumption of the grounded neutral at the substation and isolated neutral
outside the substation.
10. transformer indexed.m: This file generates the OpenDSS file for transformer
component. It incorporates currently the wye-delta configurations only as they
are the general type of transformer in the distribution system and three phase
as European Distribution system usually do not have single phase transformers.
It incorporates the idea of 4-wires LV distribution system.
11. energymeter.m: This defines an energy-meter element file for OpenDSS for
every MV line of the transformer. It provides us a chance to study the voltage
profile from that point in OpenDSS.
12. load indexed.m: This file generates an OpenDSS file for the load element of
the network. It incorporates the type of the load (single-phase or three-phase),
connected phase for single-phase load, connected bus and load shape.
73
5.5.2 Load Data
The load data from smart meter data was extracted to load shape file using this set
of codes in MATLAB listed in shortcut loadshape.m and was linked with the load
in network data file. The shape file consists of 20 days per hour data for each load
while the Loadshape file links the Load name with Load shape.
1 %shortcut_loadshape.m %it calls all program for creating loadshape file
2 days=20 %no of days of load
3 fileno=7 % no of loadfile
4 loadshape_ind; %file to Generate Loadshape file for OpenDSS
5 loadshapeextraction;%%Extracts load from the csv file and saves in vector
6 save_loadshape; %saves the extracted load in csv file
The variable ’days’ and ’fileno’ is to signify no of days in which we are interested
to extract data and number of files having the data. Since we are dealing with huge
data all load data will be in more than 1 file and we renamed all load data files by
file1, file2, etc. The different files called in the script above are discussed below:
1. loadshape ind.m: It generates an OpenDSS file to link the Load element with
the Loadshape. It gives the file name and location of respective loadshape. It
has also provision for changing number of load points in load shape.
2. loadshapeextraction.m: This file checks the smart meter data file and stores
it in a vector [shape] as per the indexing rule of Load in Load extraction module.
For example shape(1).hrs will contain 481 load points for 20 days for LOAD1
indexed in Network extraction file.
3. loadshapeextraction.m: This file is for converting the [shape] vector to .csv
file for OpenDSS program. The load shapes are saved is with name shape 1.csv,
shape 2.csv,... etc for LOAD1, LOAD2,.. as indexed in load indexed.m in
network data.
The European LV test network thus built up by using the above data extraction tool
is discussed in next chapter.
74
Chapter 6
European Standard Distribution
Network
The indexing scheme and data extraction scheme discussed in chapter 5 is now used
to generate a European LV test network representing a distribution network of the
European style.
This is the first network in the literature which includes the network data of whole
city re-annotated and representing typical European type distribution system. The set
of files would be made available publicly so that the researchers can use it for different
studies in large-scale distribution system. It has also the provisions for change in LV
Network as well as the change in the number of days of study through modification
in quasi-static load data days. Further, it includes the all type of feeders from highly
loaded to sparsely loaded and all type of transformers with sparse and dense loading.
This network includes the multiple feeders originating from a transformer as well as
reserve feeders with circuit breakers to change the network configuration. Further, it
also has circuit breakers in middle of the LV networks which can be used to modify
the load flow in Distribution Network. The load in this network is a mixture of both
3 phase and 1 phase. In this master thesis, the system has been simulated using
the existing feature in OpenDSS software but to deal with this sort of network, the
software should have a robust power flow engine like OpenDSS and a better visual
75
representation like Power World Simulator1 which can be used to operate the circuit
breakers to see the effect in load-flow. The data will also play a vital role in working
in distribution system software with better visual representation and interactive in
nature to assist distribution utility to have better visibility of network and controllable
to an extent. Further, this system is very complete to do the various economic study
as it covers a wide area with all type of load.
6.1 Introduction to Test Network
As discussed in the previous chapter, the current test feeders are mostly found in the
North American style. The existing European LV test feeder are also not complete
and does not represent actual European style feeder in terms of 4 line system with
isolated neutral, neutral grounding only in substation, different load type (1 phase
or 3 phase), integration with smart meter data and inclusion of multiple feeder in a
single transformer which has been discussed in detail in section 4.2.
With the evolution of DERs and increased penetration in the distribution network,
distribution system operators need more tools for visualization of the LV networks.
The European style network discussed in this thesis network provides the data which
can be used to test a software which supports utilities in future to control various
variables and manage DERs and storage systems more profitably. It can also be used
to test the various applications based on the smart meter data including the inclusion
of the smart meters in future SCADA making Distribution System more controllable.
The major feature of this network are:
1. The test network has LV voltage level of 416 V and frequency 50 Hz.
2. The LV network has been modeled as 4-wires system with neutral grounded
only in substation. The neutral voltage can be monitored at every bus and
neutral is isolated from consumer ground.
1https://www.powerworld.com/
76
3. The network consists of 8087 loads and 30 Distribution transformer which
represents a load of a small city. Out of 8087 loads only 31 were non-smart
meters.
4. The network consists of 10290 buses of which 2681 buses are only monitored.
5. The load is in time-series form for 20 days and every hour. The load being
obtained from smart meter directly can be modified as per need for more or
fewer days.
6.2 Network Description
The figure 6-1 shows the single line diagram of the whole network. The various
circuit elements to define the network are discussed below with the corresponding
OpenDSS file in Brackets.
Voltage Source
The source is defined by the short circuit current in three phase and 1 phase like in
IEEE European Test Feeder. But the voltage level in this system is 22 kV.
#source Impedance
edit Vsource.Source bus1=source basekv=22 pu=1.0001 phases=3 ISC3=9000
ISC1=5000
MV Lines (source2txline ind.txt)
The Imaginary MV lines are created to connect the source with the transformer.
Each MV line is assumed of very low resistance of 25×e−6Ω to neglect the MV lines.
They are three wire system and connected to the delta side of the transformer. These
lines are named mv1, mv2,...,mvn based on the number of transformers and given in
source2txline ind.txt.
#MV lines
78
New Line.mv1 Bus1=Source Bus2=1 phases=3 Linecode=101 Length=5 Units=m
New Line.mv2 Bus1=Source Bus2=2 phases=3 Linecode=101 Length=5 Units=m
New Line.mv3 Bus1=Source Bus2=3 phases=3 Linecode=101 Length=5 Units=m
Transformers (Transformer ind.txt)
The distribution transformers are in the distribution substation marked by red
delta mark in 6-1. All of the 30 transformers in network are delta-wye. The voltage
rating of the transformer are 22/0.420 kV and power rating are either 250, 630 or
1000 kVA. The reactance of the transformer winding is considered to be 4% for all
the transformer. The data of transformer is given in Transformer ind.txt of which
a part is as below:
New Transformer.TD401346 windings=2 Buses=[1 31.1.2.3.4] Conns=[Delta Wye]
kVs=[22 0.420] kVAs=[630 630] XHL=4.0 sub=y
New Transformer.TD400291 windings=2 Buses=[2 32.1.2.3.4] Conns=[Delta Wye]
kVs=[22 0.420] kVAs=[1000 1000] XHL=4.0 sub=y
New Transformer.TD000653 windings=2 Buses=[3 33.1.2.3.4] Conns=[Delta Wye]
kVs=[22 0.420] kVAs=[630 630] XHL=4.0 sub=y
LV Network (Line indexed check.txt)
The feeder and circuit breaker in the substation is defined as line element with
nomenclature as [feeder00] and [cktbk00] and is listed in Line indexed check.txt
file, which also contains the description of the all LV network. It models the LV
network as four-wires system with isolated neutral from consumer ground. The
neutral wire provides return path to the all unbalanced current. The code below
shows the representative description of the feeder, circuit breaker, and LV Networks:
!LV Feeder from Transformer to the Circuit Breaker
New Line.feeder1 Bus1=31.1.2.3.4 Bus2=31_1.1.2.3.4 phases=4 Linecode=205
Length=4.340 Units=m
79
! LV substation Circuit Breaker
New Line.cktbk1 Bus1=31_1.1.2.3.4 Bus2=2684.1.2.3.4 phases=4 Linecode=102
Length=1.000 Units=m
!LV Lines Description
New Line.67_0 Bus1=66.1.2.3.4 Bus2=67_1.1.2.3.4 phases=4 Linecode=316
Length=4.658 Units=m
New Line.67_1 Bus1=67_1.1.2.3.4 Bus2=67_2.1.2.3.4 phases=4 Linecode=316
Length=2.265 Units=m
New Line.67 Bus1=67_5.1.2.3.4 Bus2=67.1.2.3.4 phases=4 Linecode=316
Length=3.922 Units=m
The Line Codes are defined by the 4×4 matrix of R and X for LV lines and 3×3
matrix for the MV lines and is represented in OpenDSS file in linecode.txt. The
cross-section of the distribution cable is used to find the resistance and inductance.
Since the ground return-path concept is rejected in this model and the LV cables are
generally twisted insulated cables, the mutual couplings are neglected. The unique
line code is assigned for all type of cable which has been listed in A.2. The codes
below represent the way in which Line code is represented for 4-phase system.
!Line code 4x4 for LV lines
New LineCode.201 nphases=4 baseFreq=50 units=km
~ rmatrix = [1.230000 | 0 1.230000 | 0 0 1.230000 | 0 0 0 1.230000]
~ xmatrix = [ 0.080000| 0 0.080000 | 0 0 0.080000| 0 0 0 0.080000]
Load (Load indexed.txt)
The Load file is linked to the LV network by using Load indexed.txt file. There
are 8087 loads in the network which are indexed to hide the details of the consumer.
The loads are either one-phase or three-phase. In this network, the constant PQ load
with P given by load shape file and Power factor as constant 0.95 is modeled. The
80
loads are connected to the respective phase and the neutral of the distribution lines
represented by .4 which ensures that the neutral current flows back to the distribution
line neutral not ground. In this test case, 12 loads are in comments as the proper
mapping of those load was not done in GIS file and are floating without any connected
bus. This would be corrected once the utility provides the corrected connection in
GIS.
!Load Description File
New Load.LOAD1 Phases=1 Bus1=1233.3.4 kV=0.23 kW=1 PF=0.95 daily=Shape_1
New Load.LOAD382 Phases=3 Bus1=1222.1.2.3.4 kV=0.4 kW=1 PF=0.95
daily=Shape_382
The data of Loadshape is given in Loadshape.txt file and it links with the
.csv file consisting the load data for 481 hours of 20 days. Each load profiles for 20
days is given in file shape 1.csv, shape 2.csv, etc linked with LOAD1, LOAD2,
etc respectively. The load profile is for the time interval of 1 hours.
!LoadShape
New Loadshape.Shape_1 npts=481 minterval=60
mult=(file=day_20_profile\shape_1.csv) useactual=true
New Loadshape.Shape_2 npts=481 minterval=60
mult=(file=day_20_profile\shape_2.csv) useactual=true
Circuit Breaker in LV Networks (fuse indexed.txt)
The Circuit Breaker element in the middle of LV Network is the special feature of
this network. These circuit breakers are used to change the LV network configuration.
In this model, we have assumed the circuit breaker as line element with 25e−6Ω
effective resistance. To change the LV configuration, the change should be first
done in GIS representation and the extraction loop has to be run again. The visual
representation of the breaker is important in this case to understand the sequence of
operation. One of the limitations of OpenDSS is that the system cannot be simulated
81
when part of the component is disconnected from the source. So, this raw data can
be used to create a new distribution system software which can also simulate the
disconnected load without any problem. The other problem with handling fuse data
is to ensure the radial nature of the feeder for which also a better software is required
with the ability of better power flow visualization. The circuit-breaker counts are
2,4,5.. in the example below, which means that fuse1, fuse3,.. and other missing one
are circuit breakers which are open. The open circuit breakers are commented by
using ’ !’ at the beginning in the OpenDSS file.
!Circuitbreaker at Middle of LV network for changing configurations
New Line.fuse2 Bus1=2677.1.2.3.4 Bus2=1099.1.2.3.4 phases=4 Linecode=102
Length=0.5 Units=m
New Line.fuse4 Bus1=1540.1.2.3.4 Bus2=2724.1.2.3.4 phases=4 Linecode=102
Length=0.5 Units=m
New Line.fuse5 Bus1=2727.1.2.3.4 Bus2=1526.1.2.3.4 phases=4 Linecode=102
Length=0.5 Units=m
Reactor for Neutral Grounding (neutral ind.txt)
This file represents the resistance grounding of transformer neutral in TT grounding
system adopted in European distribution system. The reactor for all transformer
neutral is described in the neutral ind.txt file. Each neutral is grounded with the
resistance of 5 Ω. The .4 phase of the transformer neutral is grounded with .0 phase
which represents the earth. This grounding is required to maintain voltage balance
during extremely unbalance condition.
!Neutral Grounding of Transformer
New Reactor.grnd1 phases=1 bus1=31.4 bus2=31.0 R=5 X=0.01
New Reactor.grnd2 phases=1 bus1=32.4 bus2=32.0 R=5 X=0.01
82
Monitors
The monitor elements are used to monitor the voltage, current, active power and
reactive power of the bus 2 of the line segment monitored. This is the reason why
name of bus 2 and line segment were synchronized during data extraction. The
OpenDSS file of the network consists of three files, MVmon.txt, monitor feeder.txt
and monitor line.txt to monitor the MV side of the transformer, individual LV
feeders in the substation and all observable bus in LV network respectively. The
monitoring of MV lines will give the aggregate power of the transformer, monitoring
of LV feeders give the aggregate power of individual feeders and monitoring of the
Bus is helpful to understand the state of the bus in form of voltage, current, and
power in each phase.
!MV MOnitor
New Monitor.MV1_PQ_vs_Time Line.mv1 2 Mode=1 ppolar=0
New Monitor.MV1_VI_vs_Time Line.mv1 2 Mode=0
!LV feeder Monitor
New Monitor.feeder1_PQ_vs_Time Line.feeder1 2 Mode=1 ppolar=0
New Monitor.feeder1_VI_vs_Time Line.feeder1 2 Mode=0
!LV Line Monitor
New Monitor.LINE61_PQ_vs_Time Line.61 2 Mode=1 ppolar=0
New Monitor.LINE61_VI_vs_Time Line.61 2 Mode=0
Energymeter (energymeter ind.txt)
The energy meter is kept at MV line to keep the record of the energy consumed in
all transformers and is stored in energymeter ind.txt. It also separates the zones
of each transformer and helps to acquire the voltage and power profile of the feeders.
New energymeter.m1 LINE.mv1 1
New energymeter.m2 LINE.mv2 1
83
Bus Coordinates (Buscord indexed.txt)
The bus coordinates represented in OpenDSS format in Buscord indexed.txt has
the coordinates of all buses in the network. The file is required to represent the feeders
visually in the network plot. The text file has the bus name and X and Y coordinates.
The bus coordinate of Source is not included as the MV lines are imaginary.
!Bus Coordinates
1 284345.890000 4807588.080000
1382 284097.950000 4807649.960000
661_1 284533.640000 4807934.090000
661_2 284538.330000 4807933.180000
Master File (Master.dss)
The master file in OpenDSS is required to compile all the Distribution elements. The
Master.dss links all the individual files of different elements described above.
Clear
!! Test case
Set DefaultBaseFrequency=50 ! for European system
new circuit.test
edit Vsource.Source bus1=source basekv=22 pu=1.0001 phases=3 ISC3=9000
ISC1=5000 !Defination of Voltage Source
!LINE CODES
redirect linecode.txt !linecodes
redirect Line_indexed_check.txt !LV Lines
redirect source2txline_ind.txt !MV Lines
redirect fuse_indexed.txt !Circuit Breaker in Circuit
redirect neutral_ind.txt !Neutral Grounding
redirect Transformer_ind.txt !!Transformer Definition
solve
redirect LoadShape.txt !Load Shape Defination
84
batchedit loadshape..* useactual=no
redirect Load_indexed.txt !Load Defination
redirect monitor_feeder.txt !Feeder Monitor
redirect monitor_line.txt !LV Bus Monitor
redirect MVmon.txt !MV lines Monitor
Set Voltagebases=[22.0, 0.420]
calcv
BusCoords Buscord_indexed.txt
set casename=myckt
redirect energymeter_ind.txt
set mode=Daily number=481 stepsize=60m !Daily Power Flow Simulation for
481 hrs
solve
closedi
6.2.1 Use of MATLAB for Simulation
The Master.dss was compiled in our simulation using COM Interface in MATLAB.
The MATLAB codes consist of two files one to compile and other to observe different
bus as described below:
1. Testnetwork monitorall.m: This file compiles the Master.dss file explained
in the previous section and extracts the monitor data from OpenDSS to the
Matlab and saves it in bus.mat. The buses which are not monitored are
listed in ’notmonitored.mat’ and includes the floating buses which were
excluded during the data extraction as they are not connected to the network
in consideration. This file makes use of COM interface to use access OpenDSS
engine for Power Flow of Distribution Network.
2. plot curve.m: This file is the function to plot the various parameters of the
bus under interest after observing it in OpenDSS plot. The command structure
is plot curve(bus,busno,day1,nodays), where ’bus’ refers to the observed
85
data of all bus, ’busno’ refers to the bus under observation and ’day1’ and
’nodays’ refers to the first day under consideration and the number of days after
that respectively. The example of the output is given in B-2, which includes
active and reactive power, phase voltage and current and neutral voltage and
current.
The output file consists of bus.mat which is a structure with the bus number and
day. The day is further structured with voltage, current, active power and reactive
power for 20 days.
Figure 6-2: Output of the Power Flow Simulation of Test Network (from left)(a) Structure ’bus’ as Output (b) Expansion of structure ’bus(1443).day’ (c)Expansion of Structure ’bus(1443).day(1).Voltage’
The figure 6-2 shows the format in which the output of power flow simulation as
observed by monitors are stored in MATLAB structure. The data thus obtained can
be used to observe various parameters of a distribution network.
86
Chapter 7
Conclusions and Future Work
7.1 Conclusions
This master thesis aimed at creating a more complete test network for smart grid
applications for distribution system operator and has been able to achieve so in
following ways:
• The European LV networks being TT grounded, the 3 line model represented
through Kron’s reduction does not completely represent the actual distribution
system. So, the modified IEEE European LV test feeder is presented in chapter
4 which represents the real European feeder with unbalanced current returning
from neutral conductor instead of ground.
• With the increased penetration of smart meter and other data logging devices,
different style of time series load profile can be obtained. A new load profile with
per-second time series data was proposed in section 4.3.2 for IEEE European
test feeder. The load data was obtained from the actual household of Austria
and scaled to the consumption of the load of European test feeder.
• A sequence of code for converting the GIS data and smart meter data for
defining the network and load data in OpenDSS format was proposed in chapter
5. This module also indexes the bus and line to make it handle more easily for
the purpose of monitoring the bus. It also takes care of all the bus coordinates
87
making visual representation easier. The module further takes care of the
possible changes in LV network through the operation of the circuit breaker
in the sub-station and/or on LV network.
• In chapter 6, a real European test network is proposed which represents the
data of whole city re-annotated to easier name and hidden private data. The
network has 8087 loads and 10290 bus. The observed bus is reduced to 2681
to exclude the line segment which has almost similar power flow. This model
represents the 4-wires model with the possibility of observing the neutral voltage
and actual phase-neutral voltage obtained by the consumer. Also, it has the
real time-series data of consumers for 20 days in the interval of 1 hour.
• Many studies are being done in high voltage level to generate a synthetic test bus
with more than 70,000 bus [58, 59]. This network provides a base to compare
the similar synthetic network for the LV system of bigger cities or the whole
nation.
7.2 Future Works
The test network proposed in this thesis attempts to represent the real European
distribution system better than any other test system in the literature. But to make
it complete and robust work has to be done in few other features of software as well
as the test network.
• The current test system only simulates the LV network. The layer of MV
network can be added to it in future to represent the sub-transmission level and
also to study the effect of the change in the network configuration in the MV
network.
• The OpenDSS software provides a robust engine for power flow simulation
of a distribution network with 10,290 bus. But it lacks in term of visual
representation. The provision of multiple switches in LV network and their
88
control by the utility would be easier if the simulation software has interactive
visual programming with script based program in the base.
• The voltage and reactive power control elements are not currently used in the
test network. The test network can be used to design the software which
provides better controllability to the DSOs either in the planning stage or in
real time by the use of available reactive power control resources in LV network.
• The smart meter data can be used to analyze the LV network status in hourly
basis or even in lesser duration to solve the problem of optimal power flow in
the distribution network with more percentage of renewable generators, storage
devices, and electric vehicles.
• The test network provides a base network for various economic analysis of
future electrical distribution system in a real city as well as the study of state
estimation in distribution network by combining SCADA and smart-meter data.
• The module to create OpenDSS file can be extended to other GIS platforms for
making a universal converter.
89
Appendix A
Tables
A.1 Resistance and Impedance of Cross Section
Table A.1: Resistance and Impedance
Cross-section R(Aerial) X(A) R(UG) X(UG)
2 mm Cu 9.9 0.075 9.9 0.0754 mm Cu 4.95 0.075 4.95 0.0756 mm Cu 3.30 0.075 3.30 0.07516 mm CU 1.23 0.08 1.23 0.0816 mm Al 2.14 0.09 2.14 0.0925 mm Cu 1.34 0.097 1.538 0.09535 mm Al 0.907 0.095 0.907 0.09550 mm Al 0.718 0.093 0.718 0.09354.6 mm Al 0.658 0.09 0.658 0.0970 mm AL 0.454 0.091 0.515 0.08580 mm Al 0.39 0.090 0.450 0.08495 mm Al 0.3587 0.089 0.410 0.083150 mm Al 0.231 0.085 0.264 0.082240 mm Al 0.160 0.079 0.160 0.079Desconcido 0.210 0.075 0.21 0.075
91
A.2 Line Types and Line Code
Table A.2: Line Code for Different types of Lines used along with Circuit Breaker
LV Cable Type Aerial Underground Special
BT - MANGUERA 201 301 -BT - RV 0,6/1 KV 2*16 KAL 202 302 -BT - RV 0,6/1 KV 2*25 KAL 203 303 -BT - RV 0,6/1 KV 3(1*150 KAL) + 1*95 KAL 204 304 -BT - RV 0,6/1 KV 3(1*240 KAL) + 1*150 KAL 205 305 -BT - RV 0,6/1 KV 3(1*240 KAL) + 1*95 KAL 206 306 -BT - RV 0,6/1 KV 4*25 KAL 207 307 -BT - RV 0,6/1 KV 4*50 KAL 208 308 -BT - RV 0,6/1 KV 4*95 KAL 209 309 -BT - RX 0,6/1 KV 2*16 Cu 210 310 -BT - RX 0,6/1 KV 2*2 Cu 211 311 -BT - RX 0,6/1 KV 2*4 Cu 212 312 -BT - RX 0,6/1 KV 2*6 Cu 213 313 -BT - RZ 0,6/1 KV 2*16 AL 214 314 -BT - RZ 0,6/1 KV 3*150 AL/80 ALM 215 315 -BT - RZ 0,6/1 KV 3*150 AL/95 ALM 216 316 -BT - RZ 0,6/1 KV 3*25 AL/54,6 ALM 217 317 -BT - RZ 0,6/1 KV 3*35 AL/54,6 ALM 218 318 -BT - RZ 0,6/1 KV 3*50 AL/54,6 ALM 219 319 -BT - RZ 0,6/1 KV 3*70 ALM/54,6 AL 220 320 -BT - RZ 0,6/1 KV 3*95 AL/54,6 ALM 221 321 -BT - RZ 0,6/1 KV 4*16 AL 222 322 -BT - Desconocido BT 250 350 -HV Line - - 101LV Circuit Breaker - - 102
92
Appendix B
Figures
B.1 Isolated Element Report From OpenDSS
Figure B-1: Isolated Circuit Element Report
Note: The isolated circuit report of the OpenDSS has to be observed now to see
if there is any floating point in the OpenDSS network. These are the incorrectly
Indexed elements in GIS and should be corrected by utility in their network.
93
B.2 Example of Output by plot curve()
Figure B-2: The Output file of Power Flow simulator given byplot curve(bus,2113,1,3)
94
Bibliography
[1] Fernando E. Postigo Marcos, Carlos Mateo Domingo, Tomas GomezSan Roman, Bryan Palmintier, Bri-Mathias Hodge, Venkat Krishnan, Fernandode Cuadra Garcıa, and Barry Mather. A review of power distribution testfeeders in the united states and the need for synthetic representative networks.Energies, 10(11), 2017. ISSN 1996-1073. doi: 10.3390/en10111896. URLhttp://www.mdpi.com/1996-1073/10/11/1896.
[2] Ignacio Perez-Arriaga, C Batlle, T Gomez, J.P. Chaves, Pablo Rodilla, IgnacioHerrero, P Duenas, C.R. Vergara Ramırez, A Bharatkumar, S Burger, J Jenkins,and R Miller. Utility of the future: an mit energy initiative response to anindustry in transition, 12 2016.
[3] K. P. Schneider, B. A. Mather, B. C. Pal, C. W. Ten, G. J. Shirek, H. Zhu, J. C.Fuller, J. L. R. Pereira, L. F. Ochoa, L. R. de Araujo, R. C. Dugan, S. Matthias,S. Paudyal, T. E. McDermott, and W. Kersting. Analytic considerations anddesign basis for the ieee distribution test feeders. IEEE Transactions on PowerSystems, 33(3):3181–3188, May 2018. ISSN 0885-8950. doi: 10.1109/TPWRS.2017.2760011.
[4] W. H. Kersting. Radial distribution test feeders. IEEE Transactions on PowerSystems, 6(3):975–985, Aug 1991. ISSN 0885-8950. doi: 10.1109/59.119237.
[5] I Berganza Valmala, Gerd Bumiller, H A. Latchman, Moises Ribeiro, AlbertoSendin, E R. Wade, and L W. Yonge. Power line communications: Theory andapplications for narrowband and broadband communications over power lines,05 2010.
[6] Alberto Sendin, Txetxu Arzuaga, Iker Urrutia, Inigo Berganza, Ainara Fernan-dez, Laura Marron, Asier Llano, and Aitor Arzuaga. Adaptation of powerlinecommunications-based smart metering deployments to the requirements of smartgrids. Energies, 8(12):13481–13507, 2015. ISSN 1996-1073. doi: 10.3390/en81212372. URL http://www.mdpi.com/1996-1073/8/12/12372.
[7] W. H. Kersting. Radial distribution test feeders. In 2001 IEEE Power Engi-neering Society Winter Meeting. Conference Proceedings (Cat. No.01CH37194),volume 2, pages 908–912 vol.2, 2001. doi: 10.1109/PESW.2001.916993.
95
[8] R. C. Dugan, W. H. Kersting, S. Carneiro, R. F. Arritt, and T. E. McDermott.Roadmap for the ieee pes test feeders. In 2009 IEEE/PES Power SystemsConference and Exposition, pages 1–4, March 2009. doi: 10.1109/PSCE.2009.4840231.
[9] W. G. Sunderman, R. C. Dugan, and D. S. Dorr. The neutral-to-earth voltage(nev) test case and distribution system analysis. In 2008 IEEE Power and EnergySociety General Meeting - Conversion and Delivery of Electrical Energy in the21st Century, pages 1–6, July 2008. doi: 10.1109/PES.2008.4596812.
[10] W. H. Kersting. A three-phase unbalanced line model with grounded neutralsthrough a resistance. In 2008 IEEE Power and Energy Society General Meeting- Conversion and Delivery of Electrical Energy in the 21st Century, pages 1–2,July 2008. doi: 10.1109/PES.2008.4596760.
[11] R. F. Arritt and R. C. Dugan. The ieee 8500-node test feeder. In IEEE PES TD 2010, pages 1–6, April 2010. doi: 10.1109/TDC.2010.5484381.
[12] W. H. Kersting. A comprehensive distribution test feeder. In IEEE PES T D2010, pages 1–4, April 2010. doi: 10.1109/TDC.2010.5484418.
[13] K. Schneider, P. Phanivong, and J. S. Lacroix. Ieee 342-node low voltagenetworked test system. In 2014 IEEE PES General Meeting — ConferenceExposition, pages 1–5, July 2014. doi: 10.1109/PESGM.2014.6939794.
[14] A. A. Eajal and M. E. El-Hawary. Optimal capacitor placement and sizingin unbalanced distribution systems with harmonics consideration using particleswarm optimization. IEEE Transactions on Power Delivery, 25(3):1734–1741,July 2010. ISSN 0885-8977. doi: 10.1109/TPWRD.2009.2035425.
[15] S. Adhikari and F. Li. Coordinated v-f and p-q control of solar photovoltaicgenerators with mppt and battery storage in microgrids. IEEE Transactions onSmart Grid, 5(3):1270–1281, May 2014. ISSN 1949-3053. doi: 10.1109/TSG.2014.2301157.
[16] B. Matic-Cuka and M. Kezunovic. Islanding detection for inverter-baseddistributed generation using support vector machine method. IEEE Transactionson Smart Grid, 5(6):2676–2686, Nov 2014. ISSN 1949-3053. doi: 10.1109/TSG.2014.2338736.
[17] S. Bruno, S. Lamonaca, G. Rotondo, U. Stecchi, and M. La Scala. Unbalancedthree-phase optimal power flow for smart grids. IEEE Transactions on IndustrialElectronics, 58(10):4504–4513, Oct 2011. ISSN 0278-0046. doi: 10.1109/TIE.2011.2106099.
[18] S. A. Arefifar, Y. A. R. I. Mohamed, and T. H. M. EL-Fouly. Comprehensiveoperational planning framework for self-healing control actions in smart distri-bution grids. IEEE Transactions on Power Systems, 28(4):4192–4200, Nov 2013.ISSN 0885-8950. doi: 10.1109/TPWRS.2013.2259852.
96
[19] V. Kekatos, G. Wang, A. J. Conejo, and G. B. Giannakis. Stochastic reactivepower management in microgrids with renewables. IEEE Transactions on PowerSystems, 30(6):3386–3395, Nov 2015. ISSN 0885-8950. doi: 10.1109/TPWRS.2014.2369452.
[20] H. Hedayati, S. A. Nabaviniaki, and A. Akbarimajd. A method for placementof dg units in distribution networks. IEEE Transactions on Power Delivery, 23(3):1620–1628, July 2008. ISSN 0885-8977. doi: 10.1109/TPWRD.2007.916106.
[21] M. Nick, R. Cherkaoui, and M. Paolone. Optimal allocation of dispersed energystorage systems in active distribution networks for energy balance and gridsupport. IEEE Transactions on Power Systems, 29(5):2300–2310, Sept 2014.ISSN 0885-8950. doi: 10.1109/TPWRS.2014.2302020.
[22] E. Dall’Anese, H. Zhu, and G. B. Giannakis. Distributed optimal power flowfor smart microgrids. IEEE Transactions on Smart Grid, 4(3):1464–1475, Sept2013. ISSN 1949-3053. doi: 10.1109/TSG.2013.2248175.
[23] A. C. Rueda-Medina and A. Padilha-Feltrin. Distributed generators as providersof reactive power support - a market approach. IEEE Transactions on PowerSystems, 28(1):490–502, Feb 2013. ISSN 0885-8950. doi: 10.1109/TPWRS.2012.2202926.
[24] M. Rasheduzzaman, J. A. Mueller, and J. W. Kimball. Reduced-order small-signal model of microgrid systems. IEEE Transactions on Sustainable Energy, 6(4):1292–1305, Oct 2015. ISSN 1949-3029. doi: 10.1109/TSTE.2015.2433177.
[25] P. M. D. Oliveira-De and A. R. Quintana. New formulation for distributionsystem state estimation. In 2012 VI Andean Region International Conference,pages 3–6, Nov 2012. doi: 10.1109/Andescon.2012.11.
[26] R. R. Mojumdar, P. Arboleya, and C. Gonzalez-Moran. Step-voltage regulatormodel test system. In 2015 IEEE Power Energy Society General Meeting, pages1–5, July 2015. doi: 10.1109/PESGM.2015.7286097.
[27] W. H. Kersting. Distribution System Modeling and Analysis. 3rd ed. CRC Press,New York NY, 2012.
[28] R. Horton, W. G. Sunderman, R. F. Arritt, and R. C. Dugan. Effect ofline modeling methods on neutral-to-earth voltage analysis of multi-groundeddistribution feeders. In 2011 IEEE/PES Power Systems Conference andExposition, pages 1–6, March 2011. doi: 10.1109/PSCE.2011.5772574.
[29] A. M. Variz, J. L. R. Pereira, S. Carneiro, and P. G. Barbosa. Harmonic analysisof the power distribution neutral-to-earth voltage (nev) test case using four-wirethree-phase harmonic current injection method. In 2009 IEEE Power EnergySociety General Meeting, pages 1–7, July 2009. doi: 10.1109/PES.2009.5275758.
97
[30] K. P. Schneider, J. C. Fuller, and D. P. Chassin. Multi-state load models fordistribution system analysis. IEEE Transactions on Power Systems, 26(4):2425–2433, Nov 2011. ISSN 0885-8950. doi: 10.1109/TPWRS.2011.2132154.
[31] R. F. Arritt and R. C. Dugan. Distribution system analysis and the future smartgrid. In 2011 Rural Electric Power Conference, pages B2–1–B2–8, April 2011.doi: 10.1109/REPCON.2011.5756725.
[32] S. Zhao, X. Meng, and X. Song. Increasing maximum penetration of distributedgeneration by voltage regulation in smart distribution grid. In 2015 5thInternational Conference on Electric Utility Deregulation and Restructuring andPower Technologies (DRPT), pages 1894–1898, Nov 2015. doi: 10.1109/DRPT.2015.7432555.
[33] Zhao Yuan and Mohammad Reza Hesamzadeh. Hierarchical coordination of tso-dso economic dispatch considering large-scale integration of distributed energyresources. Applied Energy, 195:600 – 615, 2017. ISSN 0306-2619. doi: https://doi.org/10.1016/j.apenergy.2017.03.042. URL http://www.sciencedirect.
com/science/article/pii/S030626191730274X.
[34] Haibo Pen Ting Yang, Zhiyong Huang and Yajian Zhang. Optimal planning ofcommunication system of cps for distribution network. Journal of Sensors, 2017:10, 2017. doi: https://doi.org/10.1155/2017/9303989.
[35] TFWG. IEEE PES distribution systems analysis subcommittee radial testfeeders., 2016. URL http://sites.ieee.org/pes-testfeeders/.
[36] F. Ni, P. H. Nguyen, J. F. G. Cobben, H. E. van den Brom, and D. Zhao.Uncertainty analysis of aggregated smart meter data for state estimation. In2016 IEEE International Workshop on Applied Measurements for Power Systems(AMPS), pages 1–6, Sept 2016. doi: 10.1109/AMPS.2016.7602805.
[37] K. Khalid Mehmood, S. U. Khan, S. J. Lee, Z. M. Haider, M. K. Rafique, andC. H. Kim. Optimal sizing and allocation of battery energy storage systemswith wind and solar power dgs in a distribution network for voltage regulationconsidering the lifespan of batteries. IET Renewable Power Generation, 11(10):1305–1315, 2017. ISSN 1752-1416. doi: 10.1049/iet-rpg.2016.0938.
[38] A. Prostejovsky, O. Gehrke, M. Marinelli, and M. Uslar. Reduction of topologicalconnectivity information in electric power grids. In 2016 51st InternationalUniversities Power Engineering Conference (UPEC), pages 1–6, Sept 2016. doi:10.1109/UPEC.2016.8114117.
[39] W. H. Kersting. The whys of distribution system analysis. In 2010 IEEE RuralElectric Power Conference (REPC), pages A4–A4–7, May 2010. doi: 10.1109/REPCON.2010.5476210.
98
[40] W. H. Kersting and W. H. Phillips. Distribution feeder line models. IEEETransactions on Industry Applications, 31(4):715–720, Jul 1995. ISSN 0093-9994.doi: 10.1109/28.395276.
[41] EPRI. EPRI feeder J1, 2017. URL http://dpv.epri.com/feeder_j.html.
[42] I. Kim and R. G. Harley. A study on the effect of the high-penetrationphotovoltaic system on an increase in overvoltage of distribution feeders. In2015 North American Power Symposium (NAPS), pages 1–4, Oct 2015. doi:10.1109/NAPS.2015.7335204.
[43] EPRI. EPRI feeder K1, 2017. URL http://dpv.epri.com/feeder_k.html.
[44] EPRI. EPRI feeder M1, 2017. URL http://dpv.epri.com/feeder_m.html.
[45] EPRI. EPRI test circuits, 2017. URL http://svn.code.sf.net/p/
electricdss/code/trunk/Distrib/EPRITestCircuits/Readme.pdf.
[46] K. P. Schneider, Y. Chen, D. Engle, and D. Chassin. A taxonomy of northamerican radial distribution feeders. In 2009 IEEE Power Energy Society GeneralMeeting, pages 1–6, July 2009. doi: 10.1109/PES.2009.5275900.
[47] S. M. Basnet, H. Aburub, and W. Jewell. Effect of demand response on residentialenergy efficiency with direct load control and dynamic price control. In 2015IEEE Eindhoven PowerTech, pages 1–6, June 2015. doi: 10.1109/PTC.2015.7232479.
[48] S. Chouhan, D. Tiwari, H. Inan, S. Khushalani-Solanki, and A. Feliachi. Deroptimization to determine optimum bess charge/discharge schedule using linearprogramming. In 2016 IEEE Power and Energy Society General Meeting(PESGM), pages 1–5, July 2016. doi: 10.1109/PESGM.2016.7741576.
[49] Y. Li and P. J. Wolfs. Taxonomic description for western australian distributionmedium-voltage and low-voltage feeders. IET Generation, Transmission Distri-bution, 8(1):104–113, Jan 2014. ISSN 1751-8687. doi: 10.1049/iet-gtd.2013.0005.
[50] V. Rigoni, L. F. Ochoa, G. Chicco, A. Navarro-Espinosa, and T. Gozel.Representative residential lv feeders: A case study for the north west of england.IEEE Transactions on Power Systems, 31(1):348–360, Jan 2016. ISSN 0885-8950.doi: 10.1109/TPWRS.2015.2403252.
[51] V. Rigoni and A. Keane. Remote voltage estimation in lv feeders with localmonitoring at transformer level. In 2017 IEEE Power Energy Society GeneralMeeting, pages 1–5, July 2017. doi: 10.1109/PESGM.2017.8273891.
[52] T. Gomez R. Cossent F. Gangale P. Frıas C. Mateo, G. Prettico and G. Fulli.European representative electricity distribution networks. International Journalof Electrical Power and Energy Systems, 99:273 – 280, 2018. ISSN 0142-0615.doi: https://doi.org/10.1016/j.ijepes.2018.01.027.
99
[53] R. C. Dugan and T. E. McDermott. An open source platform for collaborating onsmart grid research. In 2011 IEEE Power and Energy Society General Meeting,pages 1–7, July 2011. doi: 10.1109/PES.2011.6039829.
[54] L. R. Araujo, D. R. R. Penido, S. Carneiro, J. L. R. Pereira, and P. A. N.Garcia. A comparative study on the performance of tcim full newton versusbackward-forward power flow methods for large distribution systems. In 2006IEEE PES Power Systems Conference and Exposition, pages 522–526, Oct 2006.doi: 10.1109/PSCE.2006.296368.
[55] ADREAS. Terms of use - ADRES data, 2010. URL www.ea.tuwien.ac.at.
[56] V. Arya, D. Seetharam, S. Kalyanaraman, K. Dontas, C. Pavlovski, S. Hoy,and J. R. Kalagnanam. Phase identification in smart grids. In 2011 IEEEInternational Conference on Smart Grid Communications (SmartGridComm),pages 25–30, Oct 2011. doi: 10.1109/SmartGridComm.2011.6102329.
[57] T. A. Short. Advanced metering for phase identification, transformer identifica-tion, and secondary modeling. IEEE Transactions on Smart Grid, 4(2):651–658,June 2013. ISSN 1949-3053. doi: 10.1109/TSG.2012.2219081.
[58] Adam B. Birchfield, Eran Schweitzer, Mir Hadi Athari, Ti Xu, Thomas J.Overbye, Anna Scaglione, and Zhifang Wang. A metric-based validation processto assess the realism of synthetic power grids. Energies, 10(8), 2017. ISSN 1996-1073. doi: 10.3390/en10081233. URL http://www.mdpi.com/1996-1073/10/
8/1233.
[59] A. B. Birchfield, T. Xu, and T. J. Overbye. Power flow convergence and reactivepower planning in the creation of large synthetic grids. IEEE Transactions onPower Systems, pages 1–1, 2018. ISSN 0885-8950. doi: 10.1109/TPWRS.2018.2813525.
100